From bugzilla-daemon at cs.uiuc.edu Sun Apr 1 00:45:06 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 1 Apr 2007 00:45:06 -0500 Subject: [LLVMbugs] [Bug 1298] ARM bad code: constants materialization inside loops Message-ID: <200704010545.l315j6ke006160@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1298 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |evan.cheng at apple.com, | |dalej at apple.com Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From sabre at nondot.org 2007-04-01 00:45 ------- Absolutely. We need to either do whole-function selection dags (with a code layout approach similar to the one described in cliff click's thesis) or we need to have a machine code LICM pass. -Chris *** This bug has been marked as a duplicate of 893 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 1 11:05:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 1 Apr 2007 11:05:22 -0500 Subject: [LLVMbugs] [Bug 1299] NEW: Incorrect use of RIP Relative to access globals on x86_64 Message-ID: <200704011605.l31G5MCO021219@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1299 Summary: Incorrect use of RIP Relative to access globals on x86_64 Product: libraries Version: trunk Platform: PC OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: jeffc at jolt-lang.org Absolute addressing is never used to address a global variable. This is incorrect as there is no guarantee that a given global is within 2GB of an instruction referencing it. On some operating systems, such as FreeBSD, the JIT will almost always find the globals more than 2GB away from code, but on any OS all it takes is more than 2GB of global variables to make this happen. This bug causes numerous JIT failures in llvm-test on FreeBSD. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 1 15:15:56 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 1 Apr 2007 15:15:56 -0500 Subject: [LLVMbugs] [Bug 1056] Should document gcc 3.4.4 as known-broken on x86-64 Message-ID: <200704012015.l31KFu8C026153@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1056 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |documentation Resolution| |FIXED Summary|Crash building llvm-gcc on |Should document gcc 3.4.4 as |FreeBSD x86_64 |known-broken on x86-64 ------- Additional Comments From sabre at nondot.org 2007-04-01 15:15 ------- Yes, you're right, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070326/046611.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 1 17:46:33 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 1 Apr 2007 17:46:33 -0500 Subject: [LLVMbugs] [Bug 1300] NEW: infinite loop in ARM Constant Island pass Message-ID: <200704012246.l31MkXXE029272@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1300 Summary: infinite loop in ARM Constant Island pass Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: lauro.venancio at gmail.com I think the ARM Constant Island pass algorithm doesn't converge to a solution in some rare cases. This bug makes the lencod test never finish. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 1 20:36:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 1 Apr 2007 20:36:35 -0500 Subject: [LLVMbugs] [Bug 1296] poor codegen for code size Message-ID: <200704020136.l321aZSO001505@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1296 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |code-quality Resolution| |FIXED Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-01 20:36 ------- Implemented. Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070326/046639.html Testcase here: CodeGen/Generic/phi-immediate-factoring.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 2 02:00:36 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Apr 2007 02:00:36 -0500 Subject: [LLVMbugs] [Bug 1288] Not so hot code produced for Ada array traversal Message-ID: <200704020700.l3270ad8015605@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1288 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-02 02:00 ------- The sintzero function is now optimized. Duncan is going to open other bugs for the other cases. This one testcase exposed several issues on multiple targets, yay :) -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 2 08:18:41 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Apr 2007 08:18:41 -0500 Subject: [LLVMbugs] [Bug 1301] NEW: So so code produced for Ada array traversal with a non-word index type Message-ID: <200704021318.l32DIf1j008512@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1301 Summary: So so code produced for Ada array traversal with a non- word index type Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr Split off from 1288. The attached bytecode shows the problem: these are all equivalent to a memset, or at worst to a simple array traversal setting all elements to zero, but the optimized code contains extra calculations. The ultimate origin of these is that the non-zero index value of the first array element in the original Ada code. The problem could be fixed by making the induction variable be an i32 (or i64) rather than an i8. Is this the right solution? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 2 11:55:20 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Apr 2007 11:55:20 -0500 Subject: [LLVMbugs] [Bug 1262] LTO example in the LTO docs fails to build Message-ID: <200704021655.l32GtKsd018014@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1262 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-02 11:55 ------- Done. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046666.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 2 15:32:55 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Apr 2007 15:32:55 -0500 Subject: [LLVMbugs] [Bug 1300] infinite loop in ARM Constant Island pass Message-ID: <200704022032.l32KWtta025094@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1300 dalej at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dalej at apple.com 2007-04-02 15:32 ------- Fixed here. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046683.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 2 20:18:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Apr 2007 20:18:17 -0500 Subject: [LLVMbugs] [Bug 1275] Combine zero/sign/any extended rotl/rotr patterns Message-ID: <200704030118.l331IHV4031969@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1275 scottm at aero.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 2 20:48:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 2 Apr 2007 20:48:22 -0500 Subject: [LLVMbugs] [Bug 1253] "optimized" range comparison not simplified Message-ID: <200704030148.l331mMxU000412@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1253 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|new bugs |Scalar Optimizations Keywords| |code-quality Product|new-bugs |libraries Platform|Other |All Resolution| |FIXED Target Milestone|--- |2.0 Version|unspecified |1.0 ------- Additional Comments From sabre at nondot.org 2007-04-02 20:48 ------- Fixed, testcase here: Transforms/InstCombine/xor2.ll:test[01] Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046701.html This handles the general case of a xor constant, not just this specific one. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 3 21:10:37 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 3 Apr 2007 21:10:37 -0500 Subject: [LLVMbugs] [Bug 1302] NEW: code cleanup: llvm-ld warns when linking against shared libraries Message-ID: <200704040210.l342AbEB009690@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1302 Summary: code cleanup: llvm-ld warns when linking against shared libraries Product: tools Version: trunk Platform: All OS/Version: All Status: NEW Severity: minor Priority: P2 Component: llvm-ld AssignedTo: unassignedbugs at nondot.org ReportedBy: lefever at crhc.uiuc.edu When creating a native executable with llvm-ld and linking against a shared (.so) library, using the -l option, llvm-ld generates the following warning: llvm-ld: warning: Supposed library 'name_of_lib' isn't a library. However, the produced executable seems to work fine, and seems to find libname_of_lib.so through LD_LIBRARY_PATH just fine. Reid Spencer advised that this is most likely a code cleanup bug. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 01:38:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 01:38:17 -0500 Subject: [LLVMbugs] [Bug 1302] llvm-ld warns when linking against shared libraries Message-ID: <200704040638.l346cHuB015057@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1302 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-04 01:38 ------- Fixed with these patches: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046781.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046782.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046783.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046784.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046785.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046786.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 02:28:26 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 02:28:26 -0500 Subject: [LLVMbugs] [Bug 1303] NEW: "extractelement" doesn't work for X86 Message-ID: <200704040728.l347SQnS025032@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1303 Summary: "extractelement" doesn't work for X86 Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: isanbard at gmail.com For this simple code: define i8 @bork(<16 x i8> %a) { entry: %tmp = extractelement <16 x i8> %a, i32 0 ret i8 %tmp } I get this: $ llvm-as -f b.ll $ opt -std-compile-opts -o a.bc -f b.bc $ llc -o - a.bc -march=x86 /Users/wendling/llvm/llvm.src/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:3292: failed assertion `0 && "Don't support illegal by-val vector arguments yet!"' Abort trap This doesn't abort in PPC: $ llc -o - a.bc .machine ppc7400 .section __TEXT,__textcoal_nt,coalesced,pure_instructions .section __TEXT,__symbol_stub1,symbol_stubs,pure_instructions,16 .text .globl _bork .align 4 _bork: addi r3, r1, -16 stvx v2, 0, r3 lbz r3, -16(r1) blr .subsections_via_symbols ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 02:32:32 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 02:32:32 -0500 Subject: [LLVMbugs] [Bug 1304] NEW: "extractelement" doesn't work for <1 x i64> type Message-ID: <200704040732.l347WWT1025117@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1304 Summary: "extractelement" doesn't work for <1 x i64> type Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: isanbard at gmail.com For this program: $ cat b.ll define i64 @bork(<2 x i64> %vec) { entry: %tmp = extractelement <2 x i64> %vec, i32 0 ret i64 %tmp } I get this abort: $ llvm-as -f b.ll $ opt -std-compile-opts -o a.bc -f b.bc /Users/wendling/llvm/llvm.src/lib/Transforms/Scalar/InstructionCombining.cpp:1399: failed assertion `DemandedElts != EltMask && (DemandedElts & ~EltMask) == 0 && "Invalid DemandedElts!"' Abort trap If it's more than 1 element in the vector, you get the failure reported in PR1303. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 14:17:30 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 14:17:30 -0500 Subject: [LLVMbugs] [Bug 1305] NEW: Bugpoint hard-coded to use "gcc" Message-ID: <200704041917.l34JHUvT007641@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1305 Summary: Bugpoint hard-coded to use "gcc" Product: tools Version: 1.9 Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: bugpoint AssignedTo: unassignedbugs at nondot.org ReportedBy: jeffc at jolt-lang.org A system may have more than one version of gcc installed, but bugpoint can only use the one named "gcc". By default, it ought to use the gcc with which LLVM was configured to be built with. This default may be overridden by either the CC environment variable or possibly command arguments. Furthermore, when LLVM was built with Visual Studio on Windows, it must use Microsoft C instead of gcc. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 14:42:46 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 14:42:46 -0500 Subject: [LLVMbugs] [Bug 1306] NEW: LiveVariables improperly handles killing aliased registers Message-ID: <200704041942.l34Jgke0008134@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1306 Summary: LiveVariables improperly handles killing aliased registers Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: major Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: christopher.lamb at gmail.com CC: christopher.lamb at gmail.com Here is the beginning of the BB dump. entry (0x8503c80, LLVM BB @0x8501af0, ID#0): Live Ins: %R0 %R1 %reg1024 = ORI %R0, 0 %reg1025 = ORI %R1, 0 V4R0 is getting killed because handleLiveInRegister() is called on all results of getAliasSet() for each of the liveins (this is in LiveIntervals::computeIntervals() ). handleRegisterDef() does a similar thing where calls handlePhysicalRegisterDef() on all members of getAliasSet() returned for the def, which also triggers this problem. --- This is a pretty serious bug. LiveVariables::KillsRegister should not kill aliases that are "larger". The correct way to fix this is to explicitly list registers that are defined, used, and killed. So your example should look like: entry (0x8503c80, LLVM BB @0x8501af0, ID#0): Live Ins: %R0 %R1 %reg1024 = ORI %R0, 0, %V4R0 %reg1025 = ORI %R1, 0, %V4R0 KillsRegister should check for exact match rather than regsOverlap. There are probably other similar bugs in LiveVariables. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 15:45:19 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 15:45:19 -0500 Subject: [LLVMbugs] [Bug 1307] NEW: Invalid transform due to SimplifyLibCalls Message-ID: <200704042045.l34KjJMf027454@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1307 Summary: Invalid transform due to SimplifyLibCalls Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru Consider attached .ll file. Running ./opt -simplify-libcalls will produce "bad" output. Resulting binary just outputs "Bad HeadURL". This breaks Python 2.5 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 17:27:33 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 17:27:33 -0500 Subject: [LLVMbugs] [Bug 1308] NEW: Inline asm causes assertion in llc Message-ID: <200704042227.l34MRXPe029889@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1308 Summary: Inline asm causes assertion in llc Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: jackrobi at gmail.com $ cat|llvm-as|llc >/dev/null define void @stuff(i32, ...) { %foo = alloca i8* %bar = alloca i32* %tmp10.i.i57 = call i32 asm sideeffect "movl $4,%edi", "={ax},0,i,m,m"( i32 0, i32 120, i8** %foo, i32** %bar ) ret void } ^D llc: /usr/local/code/llvm/llvm/include/llvm/CodeGen/MachineInstr.h:233: unsigned int llvm::MachineOperand::getReg() const: Assertion `isRegister() && "This is not a register operand!"' failed. llc((anonymous namespace)::PrintStackTrace()+0x19)[0x861ee09] Aborted ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 17:34:43 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 17:34:43 -0500 Subject: [LLVMbugs] [Bug 1309] NEW: store only array is never removed Message-ID: <200704042234.l34MYhmm030089@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1309 Summary: store only array is never removed Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: alenhar2 at uiuc.edu The following should be optimized: int x(int y) { int f[100]; for (y = 0; y < 100; ++y) f[y] = 0; return 1; } it generates (1.9) this unfortunate dead loop: int %x(int %y) { entry: %f = alloca [100 x int], align 16 ; <[100 x int]*> [#uses=1] br label %bb bb: ; preds = %bb, %entry %indvar = phi uint [ 0, %entry ], [ %indvar.next, %bb ] ; [#uses=2] %tmp1 = getelementptr [100 x int]* %f, int 0, uint %indvar ; [#uses=1] store int 0, int* %tmp1 %indvar.next = add uint %indvar, 1 ; [#uses=2] %exitcond = seteq uint %indvar.next, 100 ; [#uses=1] br bool %exitcond, label %bb6, label %bb bb6: ; preds = %bb ret int 1 } Thanks to Duncan and ADA for discovering this :) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 18:12:29 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 18:12:29 -0500 Subject: [LLVMbugs] [Bug 1309] store only array is never removed Message-ID: <200704042312.l34NCTaO030934@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1309 alenhar2 at uiuc.edu changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME ------- Additional Comments From alenhar2 at uiuc.edu 2007-04-04 18:12 ------- nevermind, instcombine removes this store. Yes, I'm marking my own bug as worksforme ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 20:36:41 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 20:36:41 -0500 Subject: [LLVMbugs] [Bug 1310] NEW: doesn't kill dead stores in an aggregate Message-ID: <200704050136.l351afAF001213@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1310 Summary: doesn't kill dead stores in an aggregate Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: nicholas at mxc.ca This program is a return undef: int main(void) { int foo[12][8]; foo[0][0] = 1; for (int i = 0; i < 10; ++i) { foo[i][5] = i; } foo[0][1] = 2; foo[1][1] = 3; foo[2][1] = 4; foo[3][1] = 5; return foo[4][4]; } but LLVM has unrolled the loop and otherwise left in all of the stores and the sole load. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 4 21:47:04 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 4 Apr 2007 21:47:04 -0500 Subject: [LLVMbugs] [Bug 1297] Support Overloaded Intrinsic Functions Message-ID: <200704050247.l352l4OZ002583@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1297 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-04 21:47 ------- This is now done and working with 5 intrinsics. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 5 06:20:07 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Apr 2007 06:20:07 -0500 Subject: [LLVMbugs] [Bug 1311] NEW: llvm-ld missing a dce pass Message-ID: <200704051120.l35BK7rN027635@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1311 Summary: llvm-ld missing a dce pass Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr I ran "llvm-ld -O4" on some bytecode I had to hand, and noticed the following in the result: %tmp3.i = load i8* getelementptr (%struct.system__restrictions__rident__restrictions_info* @system__restrictions__run_time_restrictions, i32 0, i32 2, i32 1) ; [#uses=1] icmp eq i8 %tmp3.i, 0 ; :0 [#uses=0] The second line is the only use of the first line, and the second line has no uses! "opt -dce" cleans this up, so presumably llvm-ld could do with an extra dce pass. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 5 11:51:31 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Apr 2007 11:51:31 -0500 Subject: [LLVMbugs] [Bug 1311] llvm-ld missing a dce pass Message-ID: <200704051651.l35GpVcw001449@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1311 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-05 11:51 ------- This should do it: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046846.html There is no testcase, so no way to verify it. Someone should eventually go through the llvm-ld passes and clean them up. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 5 21:05:43 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 5 Apr 2007 21:05:43 -0500 Subject: [LLVMbugs] [Bug 1209] Eliminate std::vector from Type Message-ID: <200704060205.l3625hMO013223@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1209 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-05 21:05 ------- Implemented with these patches: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046866.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046865.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 6 07:39:28 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Apr 2007 07:39:28 -0500 Subject: [LLVMbugs] [Bug 1309] store only array is never removed Message-ID: <200704061239.l36CdSV5001617@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1309 baldrick at free.fr changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|WORKSFORME | ------- Additional Comments From baldrick at free.fr 2007-04-06 07:39 ------- > go instcombine! is there anything it cannot do? ;-) If you add a memset, it is no longer optimized. This occurs in every optimized Ada program, since the Ada @main, which does some initialisation before calling the real "main", contains code that reduces to something almost identical to this, in which the array and a bunch of stores to it are not removed. An additional silliness in the Ada case is that the array is a zero-initialized global, and the first thing that is done to it in @main is that it is memset to 0, so the memset could be dropped. #include int x(int y) { int f[100]; memset(f,0,sizeof(f)); for (y = 0; y < 100; ++y) f[y] = 0; return 1; } ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 6 09:30:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Apr 2007 09:30:35 -0500 Subject: [LLVMbugs] [Bug 1312] NEW: llvm-upgrade rejects valid file Message-ID: <200704061430.l36EUZRL003447@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1312 Summary: llvm-upgrade rejects valid file Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llvm-upgrade AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru Consider attached file. It was generated via llvm-gcc 1.9. llvm-upgrade rejects it: ./llvm-upgrade < test.ll llvm-upgrade: :22: error: Reference to an invalid definition: 'tmp' of type 'i32' llvm-upgrade: parse failed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 6 13:43:34 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Apr 2007 13:43:34 -0500 Subject: [LLVMbugs] [Bug 1313] NEW: malloc not eliminated, even though result is only used in a null check Message-ID: <200704061843.l36IhYe1008579@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1313 Summary: malloc not eliminated, even though result is only used in a null check Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr >From the reduced example: %tmp15.i.i.i23 = malloc [2564 x i32] icmp eq [2564 x i32]* %tmp15.i.i.i23, null %retval = zext i1 %0 to i32 ret i32 %retval Believe it or not, this came up in a real program! I see no reason why the optimizers should not assume that the malloc was successful, eliminate the call and have the function return 0. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 6 13:58:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Apr 2007 13:58:22 -0500 Subject: [LLVMbugs] [Bug 1313] malloc not eliminated, even though result is only used in a null check Message-ID: <200704061858.l36IwMZi009474@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1313 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|new bugs |Scalar Optimizations Keywords| |code-quality Product|new-bugs |libraries Resolution| |FIXED Target Milestone|--- |2.0 Version|unspecified |1.0 ------- Additional Comments From sabre at nondot.org 2007-04-06 13:58 ------- Implemented, testcase here: Transforms/InstCombine/malloc2.ll Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046889.html -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 6 16:39:34 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Apr 2007 16:39:34 -0500 Subject: [LLVMbugs] [Bug 1214] llvm-gcc4 barfs on simple packed bit-fields Message-ID: <200704062139.l36LdY9D012701@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1214 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-06 16:39 ------- Yes. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 6 19:05:52 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 6 Apr 2007 19:05:52 -0500 Subject: [LLVMbugs] [Bug 1314] NEW: llc assertion failure Message-ID: <200704070005.l3705qf2015664@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1314 Summary: llc assertion failure Product: tools Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: llc AssignedTo: unassignedbugs at nondot.org ReportedBy: lefever at crhc.uiuc.edu I am getting an assertion failure when I run "llc -f -o /dev/null code.bc" code.bc has been formed by linking other bytecode files that were all generated by llvm-gcc and transformed using opt. All uses of opt passed the -verify option. I ran llc in gdb and it stops with: llc: /home/data2/lefever/work/llvm-cvs-070306/include/llvm/Support/Casting.h:199: typename llvm::cast_retty::ret_type llvm::cast(const Y&) [with X = llvm::ConstantSDNode, Y = llvm::SDOperand]: Assertion `isa(Val) && "cast() argument of incompatible type!"' failed. The stack trace at that point is: (gdb) bt #0 0x0000003de682e37d in raise () from /lib64/tls/libc.so.6 #1 0x0000003de682faae in abort () from /lib64/tls/libc.so.6 #2 0x0000003de6827c31 in __assert_fail () from /lib64/tls/libc.so.6 #3 0x000000000082829c in llvm::cast (Val=@0x166eb20) at /home/data2/lefever/work/llvm-cvs-070306/include/llvm/Support/Casting.h:199 #4 0x0000000000a6b18d in llvm::ScheduleDAG::EmitNode (this=0x1660180, Node=0x1672330, VRBaseMap=@0x7fbfffe5e0) at ScheduleDAG.cpp:543 #5 0x0000000000a6b67b in llvm::ScheduleDAG::EmitSchedule (this=0x1660180) at ScheduleDAG.cpp:616 #6 0x00000000009e4e16 in (anonymous namespace)::ScheduleDAGRRList::Schedule (this=0x1660180) at ScheduleDAGRRList.cpp:114 #7 0x0000000000a6b7e9 in llvm::ScheduleDAG::Run (this=0x1660180) at ScheduleDAG.cpp:644 #8 0x0000000000a18c1c in llvm::SelectionDAGISel::ScheduleAndEmitDAG (this=0x1011d20, DAG=@0x7fbfffec70) at SelectionDAGISel.cpp:4528 #9 0x00000000008d6052 in (anonymous namespace)::X86DAGToDAGISel::InstructionSelectBasicBlock (this=0x1011d20, DAG=@0x7fbfffec70) at X86ISelDAGToDAG.cpp:476 #10 0x0000000000a1823d in llvm::SelectionDAGISel::CodeGenAndEmitDAG (this=0x1011d20, DAG=@0x7fbfffec70) at SelectionDAGISel.cpp:4387 #11 0x0000000000a18317 in llvm::SelectionDAGISel::SelectBasicBlock (this=0x1011d20, LLVMBB=0x12780c0, MF=@0x15ee620, FuncInfo=@0x7fbfffee50) at SelectionDAGISel.cpp:4405 #12 0x0000000000a16958 in llvm::SelectionDAGISel::runOnFunction (this=0x1011d20, Fn=@0x1058130) at SelectionDAGISel.cpp:4090 #13 0x000000000098b6a9 in (anonymous namespace)::X86DAGToDAGISel::runOnFunction (this=0x1011d20, Fn=@0x1058130) at X86ISelDAGToDAG.cpp:122 #14 0x0000000000bd90b2 in llvm::FPPassManager::runOnFunction (this=0x1013020, F=@0x1058130) at PassManager.cpp:1047 #15 0x0000000000bd8e7c in llvm::FunctionPassManagerImpl::run (this=0x119be60, F=@0x1058130) at PassManager.cpp:1006 #16 0x0000000000bd8d76 in llvm::FunctionPassManager::run (this=0x7fbffff1d0, F=@0x1058130) at PassManager.cpp:951 #17 0x00000000007bd037 in main (argc=5, argv=0x7fbffff338) at llc.cpp:290 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 7 11:13:08 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 7 Apr 2007 11:13:08 -0500 Subject: [LLVMbugs] [Bug 1312] llvm-upgrade rejects valid file Message-ID: <200704071613.l37GD8kK009210@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1312 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-07 11:13 ------- Nice catch, Anton. This was a subtle bug that only showed up in long syntax call instructions. The code was using the signedness of the FunctionType ("composite") instead of the signedness of the FunctionType's result type ("signed"). Consequently it didn't find a matching (signed, "%tmp") when it looked it up because it had been stored as (composite, "%tmp"). Fortunately, an easy fix: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046950.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 7 13:55:50 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 7 Apr 2007 13:55:50 -0500 Subject: [LLVMbugs] [Bug 1291] sys::Path enhancement: add PathWithStatus Message-ID: <200704071855.l37Itoqx012739@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1291 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jeffc at jolt-lang.org Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-07 13:55 ------- Implemented with these patches: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046965.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046966.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046967.html Jeff: I think I got the Win32 part of this right, but it hasn't been compiled. Could you please try it and verify on Win32? ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 7 15:53:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 7 Apr 2007 15:53:16 -0500 Subject: [LLVMbugs] [Bug 1273] CBE fails on SingleSource/Regression/C/2003-05-22-VarSizeArray Message-ID: <200704072053.l37KrGVK015657@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1273 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |sabre at nondot.org Status|RESOLVED |REOPENED Resolution|DUPLICATE | ------- Additional Comments From rspencer at x10sys.com 2007-04-07 15:53 ------- Chris, The link in Duncan's solution is an email describing his solution and adding a patch. The description seems quite lucid and sounds like a good idea for fixing llvm-gcc pointer/array arithmetic wrt to LLVM. However, since the patch is for llvm-gcc and I couldn't commit it anyway, could you review the patch and commit or respond? Thanks, REid. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 7 16:02:00 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 7 Apr 2007 16:02:00 -0500 Subject: [LLVMbugs] [Bug 1205] LLVM IR (but not backends) should support integer bitwidths > 64 bits Message-ID: <200704072102.l37L20pa015915@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1205 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-07 16:02 ------- This work (Phase 4) has been completed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 7 16:06:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 7 Apr 2007 16:06:35 -0500 Subject: [LLVMbugs] [Bug 1261] instcombine miscompilation Message-ID: <200704072106.l37L6ZNB016070@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1261 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- BugsThisDependsOn|1205 | Status|RESOLVED |REOPENED Resolution|DUPLICATE | ------- Additional Comments From rspencer at x10sys.com 2007-04-07 16:06 ------- It seems this is a bug after all. After bug 1205 was closed (Instcombine changes to support APInt), I reran the test case. It produced: define i16 @test(i31 %zzz) { entry: %A1 = zext i31 %zzz to i32 ; [#uses=1] %B = add i32 %A1, 16384 ; [#uses=1] %C = lshr i32 %B, 15 ; [#uses=1] %D = trunc i32 %C to i16 ; [#uses=1] ret i16 %D } i.e. the "sext" changed to zext without any other compensating changes. This needs to be investigated freshly to determine if the zext is okay or if this is truly a bug. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 7 17:05:48 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 7 Apr 2007 17:05:48 -0500 Subject: [LLVMbugs] [Bug 1261] [apint] instcombine miscompilation Message-ID: <200704072205.l37M5mAC017231@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1261 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-07 17:05 ------- The Demanded Mask makes these transitions: ret i16 %D 0xFFFF (all bits demanded in %D) %D = trunc i32 %C to i16 0x0000FFFF (widen to include trunc'd bits) %C = lshr i32 %B, 15 0x7FFF8000 (get shifted bits) %B = add i32 %A, 16384 0x7FFFFFFF (16 bit value added affects low bits) %A = sext i31 %zzz to i32 0x7FFFFFFF Despite my previous comment, this looks correct to me. The high bit is not included in the shift because the truncation occurs at 16 bits while the shift was 15 bits. This is reflected in the DemandedMask shown above. In this sitaution, InstCombine opts to replace a sext with an equivalent zext because the high bit is not preserved downstream. The code looks like: // If the input sign bit is known zero, or if the NewBits are not demanded // convert this into a zero extension. if (RHSKnownZero[SrcBitWidth-1] || (NewBits & ~DemandedMask) == NewBits) { // Convert to ZExt cast CastInst *NewCast = new ZExtInst(I->getOperand(0), VTy, I->getName(), I); return UpdateValueUsesWith(I, NewCast); Presumably this is because zext is cheaper than sext on some hardware. So, I'm claiming that the sext -> zext is correct and appropriate. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 02:01:33 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 02:01:33 -0500 Subject: [LLVMbugs] [Bug 1307] Invalid transform due to SimplifyLibCalls Message-ID: <200704080701.l3871XnV003841@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1307 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-08 02:01 ------- This is now fixed, with many patches. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 08:28:37 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 08:28:37 -0500 Subject: [LLVMbugs] [Bug 1315] NEW: SimplifyLibCalls misses optimization Message-ID: <200704081328.l38DSbMX024885@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1315 Summary: SimplifyLibCalls misses optimization Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Interprocedural Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com After Chris' round of changes to SimplifyLibCalls, it is now missing an optimization that it previously performed, as identified by this test case failing: FAIL: /proj/llvm/llvm-1/test/Transforms/SimplifyLibCalls/SPrintF.ll: child process exited abnormally %r4 = call i32 (i8*, i8*, ...)* @sprintf( i8* %target_p, i8* %fmt1_p, i8* %hello_p ) ; [#uses=1] %r4.1 = call i32 (i8*, i8*, ...)* @sprintf( i8* %target_p, i8* %fmt1_p, i8* %p ) ; [#uses=1] %r5 = call i32 (i8*, i8*, ...)* @sprintf( i8* %target_p, i8* %fmt2_p, i32 82 ) ; [#uses=1] Some of these are important for performance. For example, the first one should be reduced to a few stores. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 10:19:55 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 10:19:55 -0500 Subject: [LLVMbugs] [Bug 967] [predsimplify] release run works fine, debug run goes crazy Message-ID: <200704081519.l38FJt9h027118@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=967 nicholas at mxc.ca changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From nicholas at mxc.ca 2007-04-08 10:19 ------- This is done. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 13:12:14 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 13:12:14 -0500 Subject: [LLVMbugs] [Bug 1315] SimplifyLibCalls misses optimization Message-ID: <200704081812.l38ICEdb029946@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1315 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-08 13:12 ------- Fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047003.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 17:24:10 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 17:24:10 -0500 Subject: [LLVMbugs] [Bug 1314] Inline asm crash Message-ID: <200704082224.l38MOAbM001720@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1314 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- OtherBugsDependingO| |862 nThis| | Status|NEW |RESOLVED Resolution| |FIXED Summary|ScheduleDAG::EmitNode |Inline asm crash |assertion failure | Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-08 17:24 ------- Fixed, testcase here: CodeGen/X86/2007-04-08-InlineAsmCrash.ll Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047028.html Thanks! -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 18:56:48 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 18:56:48 -0500 Subject: [LLVMbugs] [Bug 1316] NEW: Unsupported operator after Inline Asm Change Message-ID: <200704082356.l38NumM7006430@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1316 Summary: Unsupported operator after Inline Asm Change Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com Immediately after Chris' change to inline asm to fix bug 1314, this one started failing: FAIL: /proj/llvm/llvm-4/test/CodeGen/Alpha/2006-07-03-ASMFormalLowering.ll: child process exited abnormally NODE: 0x8e7f418: i32 = TargetConstant <10> llc: LegalizeDAG.cpp:3178: llvm::SDOperand::SelectionDAGLegalize::PromoteOp(llvm::SDOperand): Assertion `0 && "Do not know how to promote this operator!"' failed. .arch ev6 .set noat llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x88d235e] llc((anonymous namespace)::SignalHandler(int)+0x110)[0x88d2686] [0x7ac420] /lib/libc.so.6(abort+0x101)[0x3234f1] /lib/libc.so.6(__assert_fail+0xfd)[0x31b859] llc[0x86a9792] llc[0x86b6f95] llc[0x86b757f] llc(llvm::SelectionDAG::Legalize()+0x41)[0x86b76d1] llc(llvm::SelectionDAGISel::CodeGenAndEmitDAG(llvm::SelectionDAG&)+0x91)[0x8650b41] llc(llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*, llvm::MachineFunction&, llvm::FunctionLoweringInfo&)+0xad)[0x8664e51] llc(llvm::SelectionDAGISel::runOnFunction(llvm::Function&)+0x11f)[0x8665adf] llc(llvm::FPPassManager::runOnFunction(llvm::Function&)+0x13a)[0x885fe7a] llc(llvm::FunctionPassManagerImpl::run(llvm::Function&)+0x6e)[0x88600fe] llc(llvm::FunctionPassManager::run(llvm::Function&)+0x88)[0x88601c0] llc(main+0xbd6)[0x83b504e] /lib/libc.so.6(__libc_start_main+0xdc)[0x30f4e4] llc(__gxx_personality_v0+0x171)[0x83b3b41] Output/2006-07-03-ASMFormalLowering.ll.out.script: line 1: 28544 Done llvm-upgrade http://llvm.org/bugs/show_bug.cgi?id=1316 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-08 19:34 ------- Fixed: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047032.html This also fixes another lurking alpha inline asm bug with memory operands. Thanks, -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 19:48:01 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 19:48:01 -0500 Subject: [LLVMbugs] [Bug 1308] instrs with multiple frame indices don't get them all rewritten Message-ID: <200704090048.l390m1Ag007394@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1308 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|Backend: X86 |Common Code Generator Code Keywords| |compile-fail Resolution| |FIXED Summary|Inline asm causes assertion |instrs with multiple frame |in llc |indices don't get them all | |rewritten Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-08 19:48 ------- Fixed, testcase here: CodeGen/Generic/2007-04-08-MultipleFrameIndices.ll Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047034.html Thanks! -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 19:49:25 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 19:49:25 -0500 Subject: [LLVMbugs] [Bug 931] [llvm-gcc] ConvertStructFieldInitializerToType problems Message-ID: <200704090049.l390nPDe007435@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=931 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED Summary|ConvertStructFieldInitialize|[llvm-gcc] |rToType problems |ConvertStructFieldInitialize | |rToType problems ------- Additional Comments From sabre at nondot.org 2007-04-08 19:49 ------- Closing this as fixed. Andrew if there is a case still unhandled, please file another bug, thanks! ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 19:54:57 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 19:54:57 -0500 Subject: [LLVMbugs] [Bug 1310] Loop Unrolling thwarts dead stores elimination of aggregate elements Message-ID: <200704090054.l390svGt007566@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1310 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From sabre at nondot.org 2007-04-08 19:54 ------- The link-time optimizer should clean this up, because it does run scalarrepl after loop unroll. However, what we really need is a smarter DSE pass. I'm going to mark this as a dup of 1309. *** This bug has been marked as a duplicate of 1309 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 20:12:56 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 20:12:56 -0500 Subject: [LLVMbugs] [Bug 1286] insertelement of undef generates horrible code Message-ID: <200704090112.l391CuLX008095@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1286 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|Backend: X86 |Scalar Optimizations Keywords| |code-quality Resolution| |FIXED Summary|This SSE Code is Horrid! |insertelement of undef | |generates horrible code Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-08 20:12 ------- Fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047039.html Testcase here: Transforms/InstCombine/vec_insertelt.ll We now compile this to: _foo: movl 4(%esp), %eax movdqa (%eax), %xmm0 movl 8(%esp), %ecx psllw (%ecx), %xmm0 movdqa %xmm0, (%eax) ret GCC manages: _foo: subl $12, %esp movl 16(%esp), %edx movl 20(%esp), %eax movdqa (%eax), %xmm0 movdqa (%edx), %xmm1 psllw %xmm0, %xmm1 movdqa %xmm1, (%edx) addl $12, %esp ret -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 20:29:59 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 20:29:59 -0500 Subject: [LLVMbugs] [Bug 1303] "extractelement" doesn't work for X86 Message-ID: <200704090129.l391TxQv008514@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1303 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |INVALID ------- Additional Comments From sabre at nondot.org 2007-04-08 20:29 ------- Works for me. You need to pass -mcpu=yonah or something else that supports SSE if you want to use SSE and your current system doesn't have it (e.g. you're on a ppc). ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 20:38:53 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 20:38:53 -0500 Subject: [LLVMbugs] [Bug 1304] instcombine crash on extractelement from <1 x i64> type Message-ID: <200704090138.l391crjp008823@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1304 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|Backend: X86 |Scalar Optimizations Keywords| |compile-fail OS/Version|MacOS X |All Platform|Macintosh |All Resolution| |FIXED Summary|"extractelement" doesn't |instcombine crash on |work for <1 x i64> type |extractelement from <1 x | |i64> type Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-08 20:38 ------- Fixed, testcase here: Transforms/InstCombine/2007-04-08-SingleEltVectorCrash.ll Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047044.html Thanks, -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 21:01:04 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 21:01:04 -0500 Subject: [LLVMbugs] [Bug 1289] Drop "implementation" keyword in AsmParser Message-ID: <200704090201.l39214nc009235@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1289 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-08 21:01 ------- Its been ten days since we notified people that "implementation" was going away. Now its gone: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/047046.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 21:09:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 21:09:27 -0500 Subject: [LLVMbugs] [Bug 1192] using a ll file as source for debuging info Message-ID: <200704090209.l3929Rns009433@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1192 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |LATER ------- Additional Comments From rspencer at x10sys.com 2007-04-08 21:09 ------- Andrew, This PR has been waiting for some feedback from you. Do you still want it implemented? Is it something you need or do you have a workaround? I'm going to mark this as "LATER". Reopen if you like. Reid. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 8 21:22:12 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 8 Apr 2007 21:22:12 -0500 Subject: [LLVMbugs] [Bug 1182] predsimplify makes lamba 10% slower Message-ID: <200704090222.l392MCMK009730@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1182 nicholas at mxc.ca changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From nicholas at mxc.ca 2007-04-08 21:22 ------- I don't think my predsimplify changes helped this, maybe the backend stuff did. Either way, I've re-run the tests and performance GCC/LLC is 0.91 with and without predsimplify, so I'm closing this bug. Hopefully if I encounter something like this in the future I'll be able narrow the bug to a smaller testcase before filing. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 9 08:43:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Apr 2007 08:43:24 -0500 Subject: [LLVMbugs] [Bug 1317] NEW: cl::ParseCommandLineOptions recurses infinitely Message-ID: <200704091343.l39DhOnh004005@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1317 Summary: cl::ParseCommandLineOptions recurses infinitely Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Support Libraries AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com I'm reporting this for Zhongxing Xu . Try this: make -C lib/Transforms/Hello Debug/lib/opt -load=Debug/lib/LLVMHello.so -help It will infinitely recurse printing out things like: opt: CommandLine Error: Argument 'version' defined more than once! opt: CommandLine Error: Argument 'help-hidden' defined more than once! opt: CommandLine Error: Argument 'help' defined more than once! opt: CommandLine Error: Argument 'stats' defined more than once! opt: CommandLine Error: Argument 'info-output-file' defined more than once! opt: CommandLine Error: Argument 'track-memory' defined more than once! opt: CommandLine Error: Argument 'version' defined more than once! opt: CommandLine Error: Argument 'help-hidden' defined more than once! opt: CommandLine Error: Argument 'help' defined more than once! opt: CommandLine Error: Argument 'analyze' defined more than once! opt: CommandLine Error: Argument 'quiet' defined more than once! opt: CommandLine Error: Argument 'q' defined more than once! opt: CommandLine Error: Argument 'std-compile-opts' defined more than once! opt: CommandLine Error: Argument 'disable-opt' defined more than once! opt: CommandLine Error: Argument 'disable-inlining' defined more than once! opt: CommandLine Error: Argument 'strip-debug' defined more than once! opt: CommandLine Error: Argument 'verify-each' defined more than once! opt: CommandLine Error: Argument 'disable-verify' defined more than once! opt: CommandLine Error: Argument 'disable-output' defined more than once! opt: CommandLine Error: Argument 'p' defined more than once! opt: CommandLine Error: Argument 'f' defined more than once! opt: CommandLine Error: Argument 'o' defined more than once! opt: CommandLine Error: Argument 'disable-compression' defined more than once! opt: CommandLine Error: Argument 'printm' defined more than once! ad nasueum. Randomly breaking into the infinite loop in gdb yielded this stack trace: #5 0x0034f2de in fwrite () from /lib/libc.so.6 #6 0x0018e367 in __gnu_cxx::stdio_sync_filebuf >::xsputn (this=0x3, __s=0x3
, __n=3) at /proj/gcc/build-4.0.3/i686-pc-linux-gnu/libstdc++-v3/include/ext/stdio_sync_filebuf.h:217 #7 0x001929b1 in std::operator<< > (__out=@0x86c89c0, __s=0x86c7b60 "opt") at /proj/gcc/build-4.0.3/i686-pc-linux-gnu/libstdc++-v3/include/streambuf:425 #8 0x085d5719 in llvm::BaseStream::operator<< (this=0x86cb878, Thing=@0x86c7b60) at /proj/llvm/llvm-1/include/llvm/Support/Streams.h:44 #9 0x085d2e01 in GetOptionInfo (PositionalOpts=@0xbf923790, OptionsMap=@0xbf923778) at CommandLine.cpp:108 #10 0x085db4be in (anonymous namespace)::HelpPrinter::operator= (this=0x86cb774, Value=true) at CommandLine.cpp:895 #11 0x085dc3f1 in llvm::cl::opt_storage<(anonymous namespace)::HelpPrinter, true, true>::setValue (this=0x86cb760, V=@0xbf923837) at /proj/llvm/llvm-1/include/llvm/Support/CommandLine.h:715 #12 0x085dc44a in llvm::cl::opt<(anonymous namespace)::HelpPrinter, true, llvm::cl::parser >::handleOccurrence (this=0x86cb740, pos=2, ArgName=0xbf925634 "help", Arg=@0xbf9238d0) at /proj/llvm/llvm-1/include/llvm/Support/CommandLine.h:776 #13 0x085d2374 in llvm::cl::Option::addOccurrence (this=0x86cb740, pos=2, ArgName=0xbf925634 "help", Value=@0xbf9238d0) at CommandLine.cpp:672 #14 0x085d25c1 in ProvideOption (Handler=0x86cb740, ArgName=0xbf925634 "help", Value=0x0, argc=3, argv=0xbf923d74, i=@0xbf923998) at CommandLine.cpp:185 #15 0x085d41bf in llvm::cl::ParseCommandLineOptions (argc=@0xbf923cf0, argv=0xbf923d74, Overview=0x860be0c " llvm .bc -> .bc modular optimizer and analysis printer \n") at CommandLine.cpp:528 #16 0x083341bd in main (argc=3, argv=0xbf923d74) at opt.cpp:248 Additionally, running: Debug/lib/opt -load=Debug/lib/LLVMHello.so -hello yields: opt: Unknown command line argument '-hello'. Try: 'opt --help' ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 9 14:19:56 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Apr 2007 14:19:56 -0500 Subject: [LLVMbugs] [Bug 1317] CommandLine fails with loadable modules. Message-ID: <200704091919.l39JJu1Y010695@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1317 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED Target Milestone|--- |2.0 ------- Additional Comments From rspencer at x10sys.com 2007-04-09 14:19 ------- This should fix the instant case: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047109.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 9 14:34:10 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Apr 2007 14:34:10 -0500 Subject: [LLVMbugs] [Bug 1318] NEW: Options in loadable modules are not available Message-ID: <200704091934.l39JYAAL011001@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1318 Summary: Options in loadable modules are not available Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Support Libraries AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com This is related to bug 1317. When a loadable module is loaded its command line options are not available until the loaded module has been processed. This is a bit of a chicken-and-egg problem: 1. Option values aren't available until after ParseCommandLineOptions 2. A -load option doesn't get processed until after ParseCommandLineOptions 3. Options in the loaded module can't been seen by ParseCommandLineOptions 4. Unrecognized option errors are generated by ParseCommandLineOptions because the module hasn't been loaded yet. Consequently, you get errors about the names of optimization passes. I'm not sure how this ever worked or if its a result of the recent changes to CommandLine. There are several things we could do to rectify this: 1. Make a formal API for "foreign passes". That is, when we dlopen the shared object we look for a function of a particular name, add its address to a table of foreign passes, invoke that function to initialize, etc. This API could do many other things (like build a PassManager full of passes). 2. Incorporate the --load option into CommandLine. It would become a new type of option that contains a list of function addresses. The functions are obtained by dlopen/dlsym on the shared object. The function (in the shared object) is invoked to *properly* register any command line options. This implies that static option objects are not permitted in such objects. 3. We might want to add something to Support library that could be used (but not linked into) the loadable module. This would be an interface to make "LLVM compatible loadable modules" that takes care of many of the details. I.e. it provides a cleaner client side interface for the module writer. This lib would be linked into all tools that support --load. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 9 15:39:20 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Apr 2007 15:39:20 -0500 Subject: [LLVMbugs] [Bug 1171] Eliminate domset from LLVM Message-ID: <200704092039.l39KdKs6012177@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1171 resistor at mac.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From resistor at mac.com 2007-04-09 15:39 ------- This is, indeed, done. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 9 19:13:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 9 Apr 2007 19:13:22 -0500 Subject: [LLVMbugs] [Bug 1319] NEW: Improve failure detection in llvm/test Message-ID: <200704100013.l3A0DMAS017413@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1319 Summary: Improve failure detection in llvm/test Product: Build scripts Version: 1.6 Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Makefiles AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com Currently we are doubling up our test scripts with things like: ; RUN: llvm-as < %s | opt -loop-rotate -disable-output && ; RUN: llvm-as < %s | opt -loop-rotate | llvm-dis | not grep "\[ .tmp224" in order to ensure that any failures early in the pipeline are not causing false positives simply because no output was generated. There are two issues here: 1. Correctness. If we had a mechanism that checked each member of a pipeline for 0 return code, that would be preferable as the test case writer doesn't have to think about that sort of thing happening. By having a mechanism that does this automatically, additional test cases, not covered by the above duplication would be made more robust. 2. Speed. This solution to the problem is coming at the expense of invoking the same pipelines multiple times. While this isn't an issue now while the test suite runs quickly, it will in due time cause issues. The test suite has grown by over 1000 tests in the last year or so. Its likely to accellerate that growth rate for some time to come. This isn't a pressing need, just filing this so we don't forget about it. Reid. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 10 11:57:25 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 10 Apr 2007 11:57:25 -0500 Subject: [LLVMbugs] [Bug 1154] LICM missing hoist of calls due to missing loop rotation Message-ID: <200704101657.l3AGvPtH013790@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1154 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-10 11:57 ------- Thanks! ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 10 11:59:03 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 10 Apr 2007 11:59:03 -0500 Subject: [LLVMbugs] [Bug 1160] x86 backend misses read-modify-write opportunity Message-ID: <200704101659.l3AGx39C013851@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1160 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-10 11:59 ------- Fixed: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070402/046911.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047158.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047160.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 10 12:00:57 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 10 Apr 2007 12:00:57 -0500 Subject: [LLVMbugs] [Bug 1160] x86 backend misses read-modify-write opportunity Message-ID: <200704101700.l3AH0vB7013977@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1160 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 10 12:13:18 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 10 Apr 2007 12:13:18 -0500 Subject: [LLVMbugs] [Bug 1320] NEW: Loop Rotate does not preserve dominator info Message-ID: <200704101713.l3AHDI10014218@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1320 Summary: Loop Rotate does not preserve dominator info Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: dpatel at apple.com Update Loop Rotate pass to preserve dominator info. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 10 13:54:43 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 10 Apr 2007 13:54:43 -0500 Subject: [LLVMbugs] [Bug 1321] NEW: llvm-gcc crashes on packed bitfield in addNewBitField Message-ID: <200704101854.l3AIshHS016190@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1321 Summary: llvm-gcc crashes on packed bitfield in addNewBitField Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr testcase: struct X { unsigned int e0 : 17; unsigned int e1 : 17; unsigned int e2 : 17; unsigned int e3 : 17; unsigned int e4 : 17; unsigned int e5 : 17; unsigned int e6 : 17; unsigned int e7 : 17; } __attribute__((packed)) x; Compiling with the llvm-gcc C compiler: gcc/llvm-types.cpp:1184: void StructTypeConversionInfo::addNewBitField(unsigned int, unsigned int): Assertion `Size <= 64 && "Bitfield too large!"' failed. packed17.c:10: internal compiler error: Aborted Testcase reduced from Ada's s-pack17. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 11 10:36:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Apr 2007 10:36:27 -0500 Subject: [LLVMbugs] [Bug 1318] Options in loadable modules are not available Message-ID: <200704111536.l3BFaRF9021676@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1318 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |compile-fail, regression Resolution| |FIXED Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-11 10:36 ------- Fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047219.html sorry for the breakage! -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 11 15:03:21 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Apr 2007 15:03:21 -0500 Subject: [LLVMbugs] [Bug 1322] NEW: llvm-commits needs Reply-To field inserted Message-ID: <200704112003.l3BK3LdS029782@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1322 Summary: llvm-commits needs Reply-To field inserted Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com When replying to the llvm-commits email list, doing a reply all causes the reply to include the sender and llvm-commits email list which causes the original sender to get two copies. This is annoying. If the llvm-commits list generated a ReplyTo field, this wouldn't happen. In both ReplyAll and Reply cases only llvm-commits list would be listed ensuring the original sender only got one copy of the reply. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 11 16:15:34 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Apr 2007 16:15:34 -0500 Subject: [LLVMbugs] [Bug 1318] Options in loadable modules are not available Message-ID: <200704112115.l3BLFYWD031482@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1318 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From rspencer at x10sys.com 2007-04-11 16:15 ------- As of now, HEAD does not fix this: opt -load=Debug/lib/LLVMHello.so -hello2 opt: Unknown command line argument '-hello2'. Try: 'opt --help' Test case for this is now in test/Feature/load_module.ll Please make sure you update lib/Transforms/Makefile and build LLVMHello.so upon which this test case now depends. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 11 16:57:38 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Apr 2007 16:57:38 -0500 Subject: [LLVMbugs] [Bug 1322] llvm-commits needs Reply-To field inserted Message-ID: <200704112157.l3BLvcOL032310@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1322 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-11 16:57 ------- done ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 11 19:36:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Apr 2007 19:36:35 -0500 Subject: [LLVMbugs] [Bug 1321] llvm-gcc crashes on packed bitfield in addNewBitField Message-ID: <200704120036.l3C0aZXC003532@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1321 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-11 19:36 ------- Fixed http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047342.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 11 19:37:53 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 11 Apr 2007 19:37:53 -0500 Subject: [LLVMbugs] [Bug 1318] Options in loadable modules are not available Message-ID: <200704120037.l3C0br34003603@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1318 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-11 19:37 ------- Fixed better, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047343.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047344.html Testcase here: test/Feature/load_module.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 12 23:01:08 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 12 Apr 2007 23:01:08 -0500 Subject: [LLVMbugs] [Bug 1323] NEW: switch lowering crash Message-ID: <200704130401.l3D418Y0003580@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1323 Summary: switch lowering crash Product: libraries Version: 1.0 Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org it looks like switch lowering is creating an invalid phi node. To reproduce, use: $ llvm-as < bugpoint-reduced-simplified.ll | llc -fast This seems to only occur with -fast. Maybe switch lowering depends on critical edges being split or something? -fast is used for -O0 compiles. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 13 01:54:51 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Apr 2007 01:54:51 -0500 Subject: [LLVMbugs] [Bug 1323] switch lowering crash Message-ID: <200704130654.l3D6sp1K008070@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1323 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-04-13 01:54 ------- Fixed with http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047457.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047458.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 13 16:05:46 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Apr 2007 16:05:46 -0500 Subject: [LLVMbugs] [Bug 863] LSR should CSE expression it inserts into loops. Message-ID: <200704132105.l3DL5kkC019658@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=863 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-13 16:05 ------- Fixed, patches here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047477.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047479.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047450.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047449.html Testcases here: CodeGen/X86/isel-sink.ll, CodeGen/ARM/lsr-code-insertion.ll Here is the new BB after LSR: bb: ; preds = %bb.bb_crit_edge, %entry %iv.1 = phi i32 [ 0, %entry ], [ %iv.1.inc, %bb.bb_crit_edge ] ; [#uses=2] %iv. = phi i32 [ 0, %entry ], [ %iv..inc, %bb.bb_crit_edge ] ; [#uses=1] %tmp1 = load i32* %P3 ; [#uses=1] %tmp3 = load i32* %P4 ; [#uses=1] %tmp. = add i32 %tmp1, %tmp3 ; [#uses=1] %tmp.2 = add i32 %tmp., %iv.1 ; [#uses=2] store i32 %tmp.2, i32* %P1 store i32 %tmp.2, i32* %P2 %iv..inc = add i32 %iv., 1 ; [#uses=2] %exitcond = icmp eq i32 %iv..inc, 1000 ; [#uses=1] %iv.1.inc = add i32 %iv.1, 4 ; [#uses=1] br i1 %exitcond, label %return, label %bb.bb_crit_edge -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 13 18:17:09 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Apr 2007 18:17:09 -0500 Subject: [LLVMbugs] [Bug 1324] NEW: Reduce memory for "strings" and other simple global initializers Message-ID: <200704132317.l3DNH9Sw022406@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1324 Summary: Reduce memory for "strings" and other simple global initializers Product: libraries Version: 1.0 Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org Some programs use a large number of strings and other dense data objects. We currently represent these as a ConstantArray of ConstantInt objects. If you have a string that is 100 characters long, you're talking about ~2-3K to represent this thing, depending on how many common characters in the string there are. The absolute minimum 100 characters can take is > 1600 bytes, because that is the cost of the OperandList for the ConstantArray (doh). Way back in the early days, we had the same problem with zero initialized memory. To fix that, we introduced zeroinitializer (aka ConstantAggregateZero). I think we should introduce a new ConstantAggregateData object which can be used for arrays/structs/vectors that have intializers that don't require relocations (intializers with relocations can always use the existing ConstantArray etc). This is an extremely common case, and will cut down memory use for things like kc++ dramatically. In the case of our 100 byte string above, we'd replace a 1600 byte operand list with a 100 byte blob of memory. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 13 19:20:57 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Apr 2007 19:20:57 -0500 Subject: [LLVMbugs] [Bug 1201] Unused malloc/free don't get optimized Message-ID: <200704140020.l3E0KvMe024676@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1201 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|Miscellaneous |Scalar Optimizations |Instrumentation passes | Keywords| |code-quality Resolution| |FIXED Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-13 19:20 ------- Implemented, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047494.html Testcase here: test/Transforms/InstCombine/malloc-free-delete.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 13 21:24:30 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 13 Apr 2007 21:24:30 -0500 Subject: [LLVMbugs] [Bug 1325] NEW: miscompilation due to switch -> shift/and code Message-ID: <200704140224.l3E2OUTC027929@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1325 Summary: miscompilation due to switch -> shift/and code Product: libraries Version: 1.0 Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org Shift/And lowering is producing invalid code. I'm not sure but I think that this is the issue: LBB1_2: ;cond_next67 lbz r2, 0(r29) li r3, 1 cmplwi cr0, r2, 32 slw r3, r3, r2 bgt cr0, LBB1_4 ;bb101.preheader This allows a shift of 32. On PPC, this will produce the same value as shifting by 31. On X86, this produces the same value as shifting by 0. This is the switch in question: switch i8 %tmp69, label %cond_next67.bb101_crit_edge [ i8 32, label %cond_next67.bb37_crit_edge i8 9, label %cond_next67.bb37_crit_edge2 i8 10, label %cond_next67.bb37_crit_edge3 i8 13, label %cond_next67.bb37_crit_edge4 i8 12, label %cond_next67.bb37_crit_edge5 ] That said, it is an impressive code size win :) -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 06:18:50 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 06:18:50 -0500 Subject: [LLVMbugs] [Bug 1326] NEW: llc -enable-eh crash Message-ID: <200704141118.l3EBIo9R017554@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1326 Summary: llc -enable-eh crash Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr $ llc -enable-eh eh.bc llc: /home/duncan/tmp/llvm.master/include/llvm/CodeGen/SelectionDAGNodes.h:904: llvm::MVT::ValueType llvm::SDNode::getValueType(unsigned int) const: Assertion `ResNo < NumValues && "Illegal result number!"' failed. llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x891f23a] llc((anonymous namespace)::SignalHandler(int)+0x112)[0x891f500] [0xffffe420] /lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7ce5641] /lib/tls/i686/cmov/libc.so.6(__assert_fail+0xfb)[0xb7cdd43b] llc(llvm::SDNode::getValueType(unsigned int) const+0x39)[0x844c925] llc(llvm::SDOperand::getValueType() const+0x1d)[0x844c955] llc[0x86f5c5b] llc[0x86e9c8d] llc[0x8704eef] llc[0x870535e] llc(llvm::SelectionDAG::Legalize()+0x41)[0x87054f7] llc(llvm::SelectionDAGISel::CodeGenAndEmitDAG(llvm::SelectionDAG&)+0x93) [0x869fde9] llc(llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*, llvm::MachineFunction&, llvm::FunctionLoweringInfo&)+0xb5)[0x86b555d] llc(llvm::SelectionDAGISel::runOnFunction(llvm::Function&)+0x11f)[0x86b7173] llc[0x8576643] llc(llvm::FPPassManager::runOnFunction(llvm::Function&)+0x13a)[0x88ac6d2] llc(llvm::FunctionPassManagerImpl::run(llvm::Function&)+0x6e)[0x88ac956] llc(llvm::FunctionPassManager::run(llvm::Function&)+0x88)[0x88aca16] llc(main+0xbe9)[0x8410959] /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xdc)[0xb7ccfebc] llc[0x840ec11] Aborted (core dumped) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 07:57:19 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 07:57:19 -0500 Subject: [LLVMbugs] [Bug 1327] NEW: Recent changes to constmerge breaks Qt Message-ID: <200704141257.l3ECvJNe019275@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1327 Summary: Recent changes to constmerge breaks Qt Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru Consider the attached testscase. Running ./opt -constmerge results: opt: /home/asl/proj/llvm/src/include/llvm/ADT/ilist:120: typename bidirectional_iterator::reference llvm::ilist_iterator::operator*() const [with NodeTy = llvm::GlobalVariable]: Assertion `Traits::getNext(NodePtr) != 0 && "Dereferencing end()!"' failed. Backing out recent IPO/ConstantMerge.cpp patch fixes the problem. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 08:30:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 08:30:16 -0500 Subject: [LLVMbugs] [Bug 1325] miscompilation due to switch -> shift/and code Message-ID: <200704141330.l3EDUGNe019851@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1325 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |jeffc at jolt-lang.org Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-04-14 08:30 ------- Fixed with: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047507.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047506.html Sorry for the breakage. I wasn't be able to test things on PowerPC and I cannot understand PowerPC assembler yet :) Chris, could you please verify? I also fixed some "latent" bug, which can be only triggered on 64-bit platforms. Jeff, could you please check, whether this fixes some bugs for you or not. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 13:16:04 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 13:16:04 -0500 Subject: [LLVMbugs] [Bug 1327] Recent changes to constmerge breaks Qt Message-ID: <200704141816.l3EIG4qo025117@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1327 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-14 13:16 ------- Fixed by JeffC's patch, thanks Jeff! ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 13:42:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 13:42:27 -0500 Subject: [LLVMbugs] [Bug 1328] NEW: llvm-as aborts on no-such-intrinsic error Message-ID: <200704141842.l3EIgRUg025633@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1328 Summary: llvm-as aborts on no-such-intrinsic error Product: tools Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: llvm-as AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com For example: llvm-as: /proj/llvm/llvm-2/include/llvm/Intrinsics.gen:1208: unsigned int llvm::Function::getIntrinsicID() const: Assertion `0 && "Unknown LLVM intrinsic function!"' failed. This isn't very friendly. It should produce an error message and exit non-zero instead. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 14:05:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 14:05:24 -0500 Subject: [LLVMbugs] [Bug 1329] NEW: Memory corruption Message-ID: <200704141905.l3EJ5OY6026063@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1329 Summary: Memory corruption Product: libraries Version: trunk Platform: PC OS/Version: Windows XP Status: NEW Severity: normal Priority: P2 Component: Support Libraries AssignedTo: unassignedbugs at nondot.org ReportedBy: jeffc at jolt-lang.org While trying to debug a different problem using Visual Studio, I encountered a memory corruption that appears to go unnoticed on Unix: Debug Assertion Failed! Program: c:\llvm\win32\debug\opt.exe File: dbgdel.cpp Line: 52 Expression: _BLOCK_TYPE_IS_VALID(pHead->nBlockUse) and the stack trace is: msvcr71d.dll!operator delete(void * pUserData=0x0012f100) Line 52 + 0x51 C++ msvcr71d.dll!operator delete[](void * p=0x0012f100) Line 21 + 0x9 C++ opt.exe!llvm::SmallPtrSetImpl::~SmallPtrSetImpl() Line 69 + 0x17 C++ opt.exe!llvm::SmallPtrSet::~SmallPtrSet() + 0x2b C++ opt.exe!llvm::SmallSet::~SmallSet() + 0x2b C++ opt.exe!llvm::SetVector >,llvm::SmallSet >::~SetVector >,llvm::SmallSet >() + 0x59 C++ opt.exe!`anonymous namespace'::LCSSA::visitSubloop(llvm::Loop * L=0x00f18350) Line 118 + 0x19 C++ opt.exe!`anonymous namespace'::LCSSA::runOnFunction(llvm::Function & F={...}) Line 97 + 0x17 C++ opt.exe!llvm::FPPassManager::runOnFunction(llvm::Function & F={...}) Line 1074 + 0x15 C++ opt.exe!CGPassManager::runOnModule(llvm::Module & M={...}) Line 109 + 0x16 C++ opt.exe!llvm::MPPassManager::runOnModule(llvm::Module & M={...}) Line 1142 + 0x15 C++ opt.exe!llvm::PassManagerImpl::run(llvm::Module & M={...}) Line 1174 + 0x10 C++ opt.exe!llvm::PassManager::run(llvm::Module & M={...}) Line 1207 C++ opt.exe!main(int argc=6, char * * argv=0x00325cb8) Line 358 C++ opt.exe!mainCRTStartup() Line 398 + 0x11 C kernel32.dll!7c816fd7() ntdll.dll!7c915b4f() The command to reproduce with the attached bytecode file is: opt -std-compile-opts Output/primes.linked.rbc -o Output/primes.linked.bc -f ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 14:40:20 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 14:40:20 -0500 Subject: [LLVMbugs] [Bug 1330] NEW: llvm-upgrade fails to upgrade @llvm.memcpy intrinsic properly Message-ID: <200704141940.l3EJeK0E027002@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1330 Summary: llvm-upgrade fails to upgrade @llvm.memcpy intrinsic properly Product: tools Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: llvm-upgrade AssignedTo: rspencer at x10sys.com ReportedBy: rspencer at x10sys.com Test case: test/Transforms/InstCombine/memmove.ll The bug 1319 conversions turned this up. This test has been failing for a while without report. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 15:23:02 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 15:23:02 -0500 Subject: [LLVMbugs] [Bug 1330] llvm-upgrade fails to upgrade @llvm.memcpy intrinsic properly Message-ID: <200704142023.l3EKN24a028200@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1330 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |INVALID ------- Additional Comments From rspencer at x10sys.com 2007-04-14 15:23 ------- This is actually a bug in the test cast. The suffixed intrinsics were introduced in LLVM 1.7. llvm-upgrade only needs to upgrade from 1.9. This test case should have been updated with 1.7 but wasn't. Now it is: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047536.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 16:52:51 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 16:52:51 -0500 Subject: [LLVMbugs] [Bug 1329] Memory corruption Message-ID: <200704142152.l3ELqpVA029686@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1329 jeffc at jolt-lang.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From jeffc at jolt-lang.org 2007-04-14 16:52 ------- Fixed: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047538.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047539.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 14 16:55:28 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 14 Apr 2007 16:55:28 -0500 Subject: [LLVMbugs] [Bug 1331] NEW: Don't break TestRunner Message-ID: <200704142155.l3ELtS0B029749@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1331 Summary: Don't break TestRunner Product: Test Suite Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org With the recent tclification of dejagnu stuff, TestRunner isn't working. Please fix. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 01:33:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 01:33:22 -0500 Subject: [LLVMbugs] [Bug 1331] Make TestRunner.sh work with new Tcl exec based test harness Message-ID: <200704150633.l3F6XMHp007110@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1331 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-15 01:33 ------- This issue has been fixed. The functionality of TestRunner.sh is now embedded into the test/Makefile. It is also invokable from the top level Makefile. In either case you can run a single test with: make check-one TESTONE=path/to/test/script.ll for example: make check-one TESTONE=test/Feature/basictest.ll or even: make check-one TESTONE=Feature/basictest.ll This will run just the one script specified in an environment similar to what dejagnu provides. If the test succeeds you'll just get a "PASS: ..." line. If it fails, you'll get all the gory details. The TestRunner.sh script was modified to just run "make check-one TESTONE" appropriatly. Resolved with these patches: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047583.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047584.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047585.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047586.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047587.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 03:59:52 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 03:59:52 -0500 Subject: [LLVMbugs] [Bug 1332] NEW: Crash in addNewBitField for zero-width bitfield after padding Message-ID: <200704150859.l3F8xqUT019672@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1332 Summary: Crash in addNewBitField for zero-width bitfield after padding Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr Testcase: struct Z { int :1; int :0; int :1; } z; Result: $ gcc -c z.c cc1: llvm-types.cpp:1160: void StructTypeConversionInfo::addNewBitField(unsigned int, unsigned int): Assertion `Size <= 64 && "Bitfield too large!"' failed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 07:10:59 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 07:10:59 -0500 Subject: [LLVMbugs] [Bug 1333] NEW: loop-unswitch dies with segfault in llvm::ETNode::Below Message-ID: <200704151210.l3FCAxVx028401@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1333 Summary: loop-unswitch dies with segfault in llvm::ETNode::Below Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr opt -loop-unswitch q.1.bc WARNING: You're attempting to print out a bytecode file. This is inadvisable as it may cause display problems. If you REALLY want to taste LLVM bytecode first-hand, you can force output with the `-f' option. opt((anonymous namespace)::PrintStackTrace()+0x1a)[0x8649f76] opt((anonymous namespace)::SignalHandler(int)+0x112)[0x864a23c] [0xffffe420] opt(llvm::ETNode::DominatedBySlow(llvm::ETNode*)+0x18)[0x84001c2] opt(llvm::ETForestBase::dominates(llvm::BasicBlock*, llvm::BasicBlock*)+0xba) [0x8400282] opt(llvm::SplitCriticalEdge(llvm::TerminatorInst*, unsigned int, llvm::Pass*, bool)+0x4f0)[0x84a323e] opt[0x846652e] opt[0x84669f7] opt[0x8467f01] opt[0x846812d] opt(llvm::LPPassManager::runOnFunction(llvm::Function&)+0x3f8)[0x851d796] opt(llvm::FPPassManager::runOnFunction(llvm::Function&)+0x13a)[0x85d80be] opt(llvm::FPPassManager::runOnModule(llvm::Module&)+0x6e)[0x85d8286] opt(llvm::MPPassManager::runOnModule(llvm::Module&)+0x11d)[0x85d7d5f] opt(llvm::PassManagerImpl::run(llvm::Module&)+0x6e)[0x85d7f2c] opt(llvm::PassManager::run(llvm::Module&)+0x1a)[0x85d7f7e] opt(main+0xa00)[0x837cc70] /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xdc)[0xb7ce0ebc] opt(realloc+0x71)[0x836ec51] Segmentation fault (core dumped) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 14:13:00 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 14:13:00 -0500 Subject: [LLVMbugs] [Bug 1334] NEW: assertion failure in -loop-unroll Message-ID: <200704151913.l3FJD0Qi003234@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1334 Summary: assertion failure in -loop-unroll Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr $ opt -loop-unroll sal.bc > x.bc opt: include/llvm/Instructions.h:1079: llvm::Value* llvm::PHINode::getIncomingValue(unsigned int) const: Assertion `i*2 < getNumOperands() && "Invalid value number!"' failed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 14:41:31 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 14:41:31 -0500 Subject: [LLVMbugs] [Bug 1335] NEW: -inline crash on asm Message-ID: <200704151941.l3FJfVsu003819@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1335 Summary: -inline crash on asm Product: new-bugs Version: unspecified Platform: Other OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: baldrick at free.fr The -inline pass creates invalid bytecode from this testcase: opt -inline hi_res.bc WARNING: You're attempting to print out a bytecode file. This is inadvisable as it may cause display problems. If you REALLY want to taste LLVM bytecode first-hand, you can force output with the `-f' option. Cannot take the address of an inline asm! invoke void asm "rdtsc\0A\09movl %eax, $0\0A\09movl %edx, $1", "==imr,==imr,~{dirflag},~{fpsr},~{flags},~{dx},~{ax}"( i32* null, i32* null ) to label %.noexc unwind label %cleanup144 Broken module found, compilation aborted! ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 16:40:01 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 16:40:01 -0500 Subject: [LLVMbugs] [Bug 1335] -inline crash inlining asm into EH target Message-ID: <200704152140.l3FLe1x0006495@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1335 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|new bugs |Transformation Utilities Keywords| |compile-fail Product|new-bugs |libraries Resolution| |FIXED Summary|-inline crash on asm |-inline crash inlining asm | |into EH target Target Milestone|--- |2.0 Version|unspecified |1.9 ------- Additional Comments From sabre at nondot.org 2007-04-15 16:40 ------- A winning combination: exceptions AND inline asm! :) Fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070409/047684.html Testcase here: Transforms/Inline/2007-04-15-InlineEH.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 17:39:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 17:39:27 -0500 Subject: [LLVMbugs] [Bug 1319] Improve failure detection in llvm/test Message-ID: <200704152239.l3FMdRLd008213@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1319 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-15 17:39 ------- This has now been implemented with numerous patches. See: http://llvm.org/docs/TestingGuide.html#dgstructure for details on the new syntax for RUN: lines. All test cases needing conversion have been converted. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 15 17:48:06 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 15 Apr 2007 17:48:06 -0500 Subject: [LLVMbugs] [Bug 1336] NEW: DejaGnu regressions Message-ID: <200704152248.l3FMm6hV008398@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1336 Summary: DejaGnu regressions Product: Test Suite Version: trunk Platform: All OS/Version: All Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com As a consequence of fixing the test harness (bug 1319) we now have 31 outstanding regressions in the DejaGNU test suite. If you own any of the following areas, please fix the code or the test case. The test cases have been XFAIL'd. When you are done, please un-XFAIL your test case. These are the failing test cases: FAIL: /proj/llvm/llvm-2/test/Archive/ranlib_GNU.ll: exit(2) at line 2 FAIL: /proj/llvm/llvm-2/test/Archive/ranlib_MacOSX.ll: exit(2) at line 2 FAIL: /proj/llvm/llvm-2/test/Archive/ranlib_SVR4.ll: exit(2) at line 2 FAIL: /proj/llvm/llvm-2/test/Archive/ranlib_xpg4.ll: exit(2) at line 2 FAIL: /proj/llvm/llvm-2/test/Assembler/2002-08-19-BytecodeReader.llx: signal(SIGABRT) at line 1 FAIL: /proj/llvm/llvm-2/test/Assembler/2003-08-21-ConstantExprCast-Fold.llx: signal(SIGABRT) at line 1 FAIL: /proj/llvm/llvm-2/test/Assembler/2004-01-11-getelementptrfolding.llx: signal(SIGABRT) at line 1 FAIL: /proj/llvm/llvm-2/test/C++Frontend/2004-09-27-CompilerCrash.cpp: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/C++Frontend/2007-04-11-InlineStorageClassC++.cpp: exit(1) at line 5 FAIL: /proj/llvm/llvm-2/test/CFrontend/2002-11-07-Redefinition.c: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CFrontend/2003-07-22-ArrayAccessTypeSafety.c.tr: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CFrontend/2006-01-16-BitCountIntrinsicsUnsigned.c: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/ARM/align.ll: exit(1) at line 2 FAIL: /proj/llvm/llvm-2/test/CodeGen/ARM/hello.ll: exit(1) at line 3 FAIL: /proj/llvm/llvm-2/test/CodeGen/Alpha/cmov.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/Alpha/ctlz.ll: exit(1) at line 2 FAIL: /proj/llvm/llvm-2/test/CodeGen/Alpha/ctlz_e.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/Alpha/ctpop.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/CBackend/2004-02-15-PreexistingExternals.llx: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/CBackend/2005-07-14-NegationToMinusMinus.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/Generic/vector-constantexpr.ll: signal(SIGSEGV) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/2004-02-14-InefficientStackPointer.llx: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/fp-stack-ret.ll: exit(1) at line 4 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/fp_constant_op.llx: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/fp_load_fold.llx: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/sse-fcopysign.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/sse-load-ret.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/CodeGen/X86/vec_call.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/Transforms/GlobalDCE/2002-08-17-FunctionDGE.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/Transforms/GlobalOpt/memset.ll: exit(1) at line 1 FAIL: /proj/llvm/llvm-2/test/Transforms/IndVarsSimplify/exit_value_tests.llx: exit(1) at line 1 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 16 01:58:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Apr 2007 01:58:27 -0500 Subject: [LLVMbugs] [Bug 1328] llvm-as aborts on no-such-intrinsic error Message-ID: <200704160658.l3G6wRp8018055@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1328 rspencer at x10sys.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at x10sys.com 2007-04-16 01:58 ------- Test case here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047732.html Fixed with: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047733.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047734.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047735.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047736.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 16 02:03:26 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Apr 2007 02:03:26 -0500 Subject: [LLVMbugs] [Bug 1337] NEW: Remove duplicate (explicit) checks in test cases. Message-ID: <200704160703.l3G73QGf018169@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1337 Summary: Remove duplicate (explicit) checks in test cases. Product: Test Suite Version: trunk Platform: All OS/Version: All Status: NEW Severity: minor Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at x10sys.com Before the conversion to the Tcl exec based test harness, many tests were duplicating the script to ensure that it ran correctly before actually processing its results. For example: ; RUN: llvm-as < %s | opt -blah | llvm-dis -o /dev/null -f && ; RUN: llvm-as < %s | opt -blah | llvm-dis -o - | not grep bitcast This is no longer necessary as each component in each pipeline is now checked by Tcl. Additionally the test harness enforces success of each line of RUN commands as if there was an && at the end of each line. Consequently, the duplication of commands by these test cases is merely a waste of CPU time. They should be removed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 16 16:25:10 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Apr 2007 16:25:10 -0500 Subject: [LLVMbugs] [Bug 1274] [llvm-gcc] front-end crash in lower_nested_functions, not llvm-related Message-ID: <200704162125.l3GLPAtE016472@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1274 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |LATER ------- Additional Comments From sabre at nondot.org 2007-04-16 16:25 ------- Agreed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 16 16:33:21 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Apr 2007 16:33:21 -0500 Subject: [LLVMbugs] [Bug 36] PassManager enhancements Message-ID: <200704162133.l3GLXLEa016872@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=36 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 16 18:04:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Apr 2007 18:04:35 -0500 Subject: [LLVMbugs] [Bug 1334] assertion failure in -loop-unroll Message-ID: <200704162304.l3GN4ZRc019561@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1334 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-16 18:04 ------- Fixed http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047889.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 16 19:49:23 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 16 Apr 2007 19:49:23 -0500 Subject: [LLVMbugs] [Bug 1338] NEW: When SVN lands, we should rename... Message-ID: <200704170049.l3H0nNi8021714@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1338 Summary: When SVN lands, we should rename... Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org include/llvm/ADT/ilist and friends to end with ".h". Probably others, they should be listed here. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 04:22:03 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 04:22:03 -0500 Subject: [LLVMbugs] [Bug 331] [X86/win32] Large allocas need incremental stack probing Message-ID: <200704170922.l3H9M3eY007993@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=331 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-04-17 04:22 ------- Fixed here (both issues): http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047911.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047912.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047913.html Testcase here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070416/047910.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 13:43:06 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 13:43:06 -0500 Subject: [LLVMbugs] [Bug 1339] NEW: dollar-signs in identifiers in assembly not protected Message-ID: <200704171843.l3HIh6O2020909@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1339 Summary: dollar-signs in identifiers in assembly not protected Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: djg at cray.com In this testcase: declare void @"$foo"() define void @bar() { call void @"$foo"() ret void } For x86, llvm emits the following: call $foo With GAS, this gets an error: test.s:9: Error: suffix or operands invalid for `call' This can be fixed by enclosing the name in parentheses, like this: call ($foo) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 14:15:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 14:15:24 -0500 Subject: [LLVMbugs] [Bug 1340] NEW: llc: strange message and segfault Message-ID: <200704171915.l3HJFO8v021991@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1340 Summary: llc: strange message and segfault Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonigerber at gmx.net llc emits some messages and then segfaults on the attached .bc file. Reproduce with: $ llc test.bc llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x855feba] llc(llvm::MachineFunctionPass::runOnFunction(llvm::Function&)+0x29)[0x81011c9] Segmentation fault I'm using llvm-1.9 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 14:16:39 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 14:16:39 -0500 Subject: [LLVMbugs] [Bug 1341] NEW: llc: strange message and segfault Message-ID: <200704171916.l3HJGd1h022041@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1341 Summary: llc: strange message and segfault Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: tonigerber at gmx.net llc emits some messages and then segfaults on the attached .bc file. Reproduce with: $ llc test.bc llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x855feba] llc(llvm::MachineFunctionPass::runOnFunction(llvm::Function&)+0x29)[0x81011c9] Segmentation fault I'm using llvm-1.9 ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 14:21:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 14:21:17 -0500 Subject: [LLVMbugs] [Bug 1341] llc: strange message and segfault Message-ID: <200704171921.l3HJLHXc022155@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1341 tonigerber at gmx.net changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From tonigerber at gmx.net 2007-04-17 14:21 ------- Crap. I pressed refresh in my browser and it seems it re-submitted the bug #1340 as a new one (sorry, wouldn't have expected that). Please delete. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 15:15:28 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 15:15:28 -0500 Subject: [LLVMbugs] [Bug 740] Update FAQ Message-ID: <200704172015.l3HKFSvG023111@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=740 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME ------- Additional Comments From sabre at nondot.org 2007-04-17 15:15 ------- This is a continual process, I don't think a bug is needed for this. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 17 17:59:25 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 17 Apr 2007 17:59:25 -0500 Subject: [LLVMbugs] [Bug 388] Put the Core LLVM classes on a diet Message-ID: <200704172259.l3HMxPR9026443@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=388 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-17 17:59 ------- The IR datastructures have shrunk a lot since the last update to this bug. There are a ton of ideas in this bug, here are ones that are left: 5. AbstractTypeUsers should be very few. It might be worthwhile to investigate turning Type::AbstractTypeUsers into an slist (instead of a vector) saving 8 bytes in DerivedTypes. I don't know if this is still really very interesting. As such, I'm going to close this. For specific ideas, we can either just do them or file bugs for specific improvements (like Bug 889). -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 18 13:25:02 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 18 Apr 2007 13:25:02 -0500 Subject: [LLVMbugs] [Bug 1342] NEW: llvm-gcc bootstrapping failed on PPC Message-ID: <200704181825.l3IIP2Uo026064@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1342 Summary: llvm-gcc bootstrapping failed on PPC Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: blocker Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: evan.cheng at apple.com Stage1 crashes on gcc/cp/cxx-pretty-print.c ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 19 23:03:58 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 19 Apr 2007 23:03:58 -0500 Subject: [LLVMbugs] [Bug 1343] NEW: Direct code generation of function level inline assembly Message-ID: <200704200403.l3K43wsK023686@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1343 Summary: Direct code generation of function level inline assembly Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: christopher.lamb at gmail.com ReportedBy: christopher.lamb at gmail.com It should be possible to emit machine code for inline assembly at function scope. A starting point would be supporting only "simple" inline assembly, that is it contains only a single instruction, and then build up complexity as needed from there. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 21 03:51:34 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 21 Apr 2007 03:51:34 -0500 Subject: [LLVMbugs] [Bug 400] Add support to LLVM for alignment specification of memory objects Message-ID: <200704210851.l3L8pYvL012525@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=400 christopher.lamb at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 21 14:21:32 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 21 Apr 2007 14:21:32 -0500 Subject: [LLVMbugs] [Bug 1344] NEW: CFrontend/2003-10-09-UnionInitializerBug.c fails on x86_64 Message-ID: <200704211921.l3LJLWvi023498@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1344 Summary: CFrontend/2003-10-09-UnionInitializerBug.c fails on x86_64 Product: new-bugs Version: unspecified Platform: PC OS/Version: FreeBSD Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: jeffc at jolt-lang.org Probably due to different alignments on x86_64 vs. x86. FAIL: /usr/home/jeffc/llvm/test/CFrontend/2003-10-09-UnionInitializerBug.c Failed with exit(1) at line 1 while running: /home/jeffc/llvm-gcc/install/bin/llvm-gcc -emit-llvm -S /usr/home/jeffc/llvm/test/CFrontend/2003-10-09-UnionInitializerBug.c -o - | llvm-as -f -o /dev/null Assertion failed: (TD.getABITypeAlignment(FieldTy) >= TD.getABITypeAlignment(Val->getType()) && "Field initialize is over aligned for LLVM type!"), function ConvertStructFieldInitializerToType, file ../../gcc/llvm-convert.cpp, line 5598. /usr/home/jeffc/llvm/test/CFrontend/2003-10-09-UnionInitializerBug.c:16: internal compiler error: Abort trap: 6 Please submit a full bug report, with preprocessed source if appropriate. See for instructions. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 21 15:52:49 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 21 Apr 2007 15:52:49 -0500 Subject: [LLVMbugs] [Bug 400] Add support to LLVM for alignment specification of memory objects Message-ID: <200704212052.l3LKqnZZ025116@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=400 christopher.lamb at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From christopher.lamb at gmail.com 2007-04-21 15:52 ------- It appears that vector loads don't currently have a subclass of SDNode, which is where alignment information for other loads/stores is kept. This will need to be fixed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 22 18:17:54 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 22 Apr 2007 18:17:54 -0500 Subject: [LLVMbugs] [Bug 400] Add support to LLVM for alignment specification of memory objects Message-ID: <200704222317.l3MNHsEt005743@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=400 christopher.lamb at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From christopher.lamb at gmail.com 2007-04-22 18:17 ------- The patches have been fixed and reapplied. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 23 12:08:42 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Apr 2007 12:08:42 -0500 Subject: [LLVMbugs] [Bug 1345] NEW: instcombine should recover more type info Message-ID: <200704231708.l3NH8gB6002717@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1345 Summary: instcombine should recover more type info Product: libraries Version: trunk Platform: Macintosh OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Scalar Optimizations AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org Instcombine should eliminate the two bitcasts in this example: define i8* @test(i8* %v) { %A = alloca [4 x i8*], align 16 ; <[4 x i8*]*> [#uses=3] %B = getelementptr [4 x i8*]* %A, i32 0, i32 0 ; [#uses=1] store i8* null, i8** %B %C = bitcast [4 x i8*]* %A to { [16 x i8] }* ; <{ [16 x i8] }*> [#uses=1] %D = getelementptr { [16 x i8] }* %C, i32 0, i32 0, i32 8 ; [#uses=1] %E = bitcast i8* %D to i8** ; [#uses=1] store i8* %v, i8** %E %F = getelementptr [4 x i8*]* %A, i32 0, i32 2 ; [#uses=1] %G = load i8** %F ; [#uses=1] ret i8* %G } Doing so would allow SROA to eliminate the alloca, turning the function into an identity function. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 23 15:26:51 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Apr 2007 15:26:51 -0500 Subject: [LLVMbugs] [Bug 1346] NEW: llvm-gcc unable to handle NOT operation Message-ID: <200704232026.l3NKQpDw008431@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1346 Summary: llvm-gcc unable to handle NOT operation Product: tools Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: dpatel at apple.com --- r.c --- extern bar(void *); void f(void *cd) { bar(((void *)((unsigned long)(cd) ^ -1))); } --- $ llvm-gcc /tmp/r.c -c /tmp/r.c: In function ???f???: /tmp/r.c:4: internal compiler error: Bus error Please submit a full bug report, with preprocessed source if appropriate. See for instructions $ ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 23 15:29:44 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Apr 2007 15:29:44 -0500 Subject: [LLVMbugs] [Bug 1347] NEW: llvm/test are not run using -m64 Message-ID: <200704232029.l3NKTiKj008488@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1347 Summary: llvm/test are not run using -m64 Product: Test Suite Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: dpatel at apple.com Follow these steps: $ cd llvm $ CFLAGS="-m64" ./configure $ CFLAGS="-m64" make $ cd test $ make RUNTESTFLAGS="CFrontEnd/dg.exp" llvm-gcc tests are not run using -m64, as I expected. Same is true for SingleSource tests. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 23 17:09:21 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Apr 2007 17:09:21 -0500 Subject: [LLVMbugs] [Bug 1344] CFrontend/2003-10-09-UnionInitializerBug.c fails on x86_64 Message-ID: <200704232209.l3NM9L3h010845@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1344 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-23 17:09 ------- Fixed. http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048332.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 23 19:02:00 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 23 Apr 2007 19:02:00 -0500 Subject: [LLVMbugs] [Bug 1348] NEW: x86_64 codegen bug in md5.c Message-ID: <200704240002.l3O020pS015264@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1348 Summary: x86_64 codegen bug in md5.c Product: libraries Version: trunk Platform: Macintosh OS/Version: Linux Status: NEW Severity: blocker Priority: P2 Component: Backend: X86 AssignedTo: unassignedbugs at nondot.org ReportedBy: scottm at aero.org This occurs while compiling the C frontend, xgcc produces the following diagnostic: (bosbok) libiberty make if [ x"" != x ]; then \ /work/scottm/llvm-cfrontend/obj/x86_64-unknown-linux-gnu/gcc/xgcc -B/work/scottm/llvm-cfrontend/obj/x86_64-unknown-linux-gnu/gcc/ -B/work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/bin/ -B/work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/lib/ -isystem /work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/include -isystem /work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/sys-include -c -DHAVE_CONFIG_H -O2 -g -O2 -I. -I../../../../libiberty/../include -W -Wall -Wtraditional -pedantic ../../../../libiberty/md5.c -o pic/md5.o; \ else true; fi /work/scottm/llvm-cfrontend/obj/x86_64-unknown-linux-gnu/gcc/xgcc -B/work/scottm/llvm-cfrontend/obj/x86_64-unknown-linux-gnu/gcc/ -B/work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/bin/ -B/work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/lib/ -isystem /work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/include -isystem /work/scottm/llvm-cfrontend/i686-unknown-linux-gnu/x86_64-unknown-linux-gnu/sys-include -c -DHAVE_CONFIG_H -O2 -g -O2 -I. -I../../../../libiberty/../include -W -Wall -Wtraditional -pedantic ../../../../libiberty/md5.c -o md5.o /tmp/ccSlUbtP.s: Assembler messages: /tmp/ccSlUbtP.s:924: Error: suffix or operands invalid for `sub' The relevant portion from the assembler source is: .align 16 .globl md5_buffer .type md5_buffer, at function md5_buffer: subq $4294966344, %rsp movq %rbp, -960(%rsp) leaq -960(%rsp), %rbp movq %rbx, -8(%rbp) movq %r12, -16(%rbp) movq %r14, -24(%rbp) movq %r15, -32(%rbp) The "$42949663444" is the invalid constant about which xgcc (actually, gas) complains. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 00:51:50 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 00:51:50 -0500 Subject: [LLVMbugs] [Bug 1349] NEW: GenCodeEmitter to associate MI operand number by name rather than declaration order Message-ID: <200704240551.l3O5poqk027338@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1349 Summary: GenCodeEmitter to associate MI operand number by name rather than declaration order Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: christopher.lamb at gmail.com The GenCodeEmitter tablegen back end should look up MI operand number by name, rather than assuming the order of declarations matches the order of the operands. This way tablegen can emit an error if the field name and operand name fail to match rather than silently producing an incorrect GenCodeEmitter. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 00:59:12 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 00:59:12 -0500 Subject: [LLVMbugs] [Bug 1350] NEW: Vreg subregs support Message-ID: <200704240559.l3O5xCo2027466@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1350 Summary: Vreg subregs support Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: christopher.lamb at gmail.com >From the dev mailing list: The issue I'm having is that there is no extract/insert instruction in the ISA, it's simply based on using subregister operands in subsequent/preliminary instructions. At the pointer of custom lowering register allocation has not yet been done, so I don't have a way to communicate the dependency. Ok. If I have a register v4r0 with subregisters {r0, r1, r2, r3} and a DAG that looks like load v4si <- extract_element 2 <- add -> load i32 I'd like to be able to generate load v4r0 load r10 add r11, r10, r2 <== subregister 2 of v4r0 Nice ISA. That is entirely too logical. :) We have a similar problem on X86. In particular, an integer truncate or an extend (e.g. i16 -> i8) wants to make use of subregisters. Consider code like this: t1 = load i16 t2 = truncate i16 t1 to i8 t3 = add i8 t2, 42 What we would really want to generate is something like this at the machine instr level: r1024 = X86_LOADi16 ... ;; r1024 is i16 r1026 = ADDi8 r1024[subreg #0], 42 More specifically, we want to be able to define, for each register class, a set of subregister classes. In the X86 world, the 64-bit register classes could have subregclass0 = i8 parts, subregclass1 = i16 parts, subregclass2 = i32 parts. Each pair should map to another physreg (e.g. -> AX). The idea of this is that the register allocator allocates registers like normal, but when it does the rewriting pass, when it replaces vregs with pregs (e.g. r1024 with CX in this example), it rewrites r1024[subreg0] with CL instead of CX. This would give us this code: CX = X86_LOADi16 ... DL = ADDi8 CL, 42 In your case, you'd define your vector register class with 4 subregs, one for each piece. Unfortunately, none of this exists yet :(. To handle truncates and extends on X86, we currently emulate this by generating machineinstrs like: r1024 = X86_LOADi16 ... r1025 = TRUNCATE_i16_to_i8 r1024 r1026 = ADDi8 r1025, 42 In the asmprinter, we print TRUNCATE_i16_to_i8 as a commented out noop if the register allocator happens to allocate 1024 and 1025 to the same register. If not, it uses an asmprinter hack to print this as a copy instruction. This is horrible, and doesn't produce good code. OTOH, before Evan improved this, we always copied into AX and out of AL for each i16->i8 truncate, which was much worse :) I see that Evan has added getSubRegisters()/getSuperRegisters() to MRegisterInfo. This is what's needed in order to implement the register allocation constraint, but there's no way yet to pass the constraint through the operands from the DAG. There would need to be some way to specify that the SDOperand is referencing a subvalue of the produced value (perhaps a subclass of SDOperand?). This would allow the register allocator to try to use the sub/super register sets to perform the instert/extract. Right. Evan is currently focusing on getting the late stages of the code generator (e.g. livevars) to be able to understand arbitrary machine instrs in the face of physreg subregs. This lays the groundwork for handling vreg subregs, but won't solve it directly. Is any of this kind of work planned? The addition of those MRegisterInfo functions has me curious... This is on our mid-term plan, which means we'll probably tackle it over the next year or so, but we don't have any concrete plans in the immediate future. If you are interested, this should be a pretty reasonable project that will give you a chance to become more familiar with various pieces of the early code generator. :) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 02:59:43 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 02:59:43 -0500 Subject: [LLVMbugs] [Bug 1351] NEW: Invalid Operand Found in Inline ASM Message-ID: <200704240759.l3O7xhS2006358@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1351 Summary: Invalid Operand Found in Inline ASM Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Backend: PowerPC AssignedTo: unassignedbugs at nondot.org ReportedBy: isanbard at gmail.com This program: define void @__gmpn_divrem_1(i32 %d) { entry: %tmp14 = icmp slt i32 %d, 0 ; [#uses=1] br i1 %tmp14, label %cond_next408, label %cond_false701 cond_next408: ; preds = %entry %tmp553 = icmp sgt i32 0, -1 ; [#uses=1] br i1 %tmp553, label %bb419, label %bb692 bb419: ; preds = %cond_next408 %tmp531 = call i32 asm "$(a${5:I}$|add${5:I}c$) $1,$4,$5\0A\09$(ae$|adde$) $0,$2,$3", "=r,==r,r,r,r,rI"( i32 * null, i32 0, i32 0, i32 0, i32 0 ) ; [#uses=0] unreachable bb692: ; preds = %cond_next408 ret void cond_false701: ; preds = %entry ret void } Results in this error: Invalid operand found in inline asm: '$(a${5:I}$|add${5:I}c$) $1,$4,$5 $(ae$|adde$) $0,$2,$3' INLINEASM , 10, %r2, 10, %r3, 9, %r2, 9, %r2, 9, % r2, 9, %r2 INLINEASM 10 %mreg(57) 10 %mreg(68) 9 %mreg(57) 9 %mreg(57) 9 %mreg(57) 9 %mreg(57) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 11:24:03 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 11:24:03 -0500 Subject: [LLVMbugs] [Bug 984] test/C++Frontend/2003-11-02-WeakLinkage.cpp.tr xfailed Message-ID: <200704241624.l3OGO36w006546@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=984 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-24 11:24 ------- Dale fixed this. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 11:33:30 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 11:33:30 -0500 Subject: [LLVMbugs] [Bug 1309] DSE is not aggressive enough Message-ID: <200704241633.l3OGXUSn006757@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1309 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-24 11:33 ------- The original testcase is fixed. I don't think this bug should continue to be open as long as there is ever some missed DSE opportunity :) -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 15:34:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 15:34:24 -0500 Subject: [LLVMbugs] [Bug 1352] NEW: volatile lost in struct copy Message-ID: <200704242034.l3OKYOVu013584@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1352 Summary: volatile lost in struct copy Product: tools Version: 1.9 Platform: All OS/Version: All Status: NEW Severity: minor Priority: P2 Component: llvm-gcc AssignedTo: unassignedbugs at nondot.org ReportedBy: djg at cray.com I typed this into the online demo: struct foo { int x; }; void copy(volatile struct foo *p, struct foo *q) { *p = *q; } It produced this: void %copy(%struct.foo* %p, %struct.foo* %q) { entry: %tmp2 = getelementptr %struct.foo* %q, int 0, uint 0 ; [#uses=1] %tmp = load int* %tmp2 ; [#uses=1] %tmp4 = getelementptr %struct.foo* %p, int 0, uint 0 ; [#uses=1] store int %tmp, int* %tmp4 ret void } The store should be volatile. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 16:32:03 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 16:32:03 -0500 Subject: [LLVMbugs] [Bug 1260] Need MMX support for Qt operations Message-ID: <200704242132.l3OLW31m015302@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1260 isanbard at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 17:29:08 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 17:29:08 -0500 Subject: [LLVMbugs] [Bug 1353] NEW: Code generator needs DominatorTree Message-ID: <200704242229.l3OMT8Aw016836@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1353 Summary: Code generator needs DominatorTree Product: libraries Version: trunk Platform: Macintosh OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org Various bits of the code generator need dominator information. The register allocator wants loop depth to compute spill weight, and future phi elimination stuff need dominator information. Currently we look back at the LLVM CFG, but this doesn't work well for multiple reasons (there is a many to one mapping from llvm bb to MBBs, MBB edges can get deleted, etc). -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 17:51:48 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 17:51:48 -0500 Subject: [LLVMbugs] [Bug 1351] PPC inline asm: "I" modifier character not supportetd Message-ID: <200704242251.l3OMpmoZ017357@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1351 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |compile-fail Resolution| |FIXED Summary|Invalid Operand Found in |PPC inline asm: "I" modifier |Inline ASM |character not supportetd Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-24 17:51 ------- Fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048368.html Testcase here: CodeGen/PowerPC/2007-04-24-InlineAsm-I-Modifier.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 19:26:49 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 19:26:49 -0500 Subject: [LLVMbugs] [Bug 1352] [llvm-gcc] volatile lost in struct copy Message-ID: <200704250026.l3P0QnIc019369@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1352 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED Summary|volatile lost in struct copy|[llvm-gcc] volatile lost in | |struct copy Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-24 19:26 ------- Fixed. Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048374.html Testcase here: test/CFrontend/2007-04-24-VolatileStructCopy.c Thanks! -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 19:42:24 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 19:42:24 -0500 Subject: [LLVMbugs] [Bug 1340] llc: strange message and segfault Message-ID: <200704250042.l3P0gOjO019766@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1340 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME ------- Additional Comments From sabre at nondot.org 2007-04-24 19:42 ------- Looks like this is fixed in LLVM 2.0. I get this output: main: subl $2, %esp fnstcw (%esp) movb $2, 1(%esp) fldcw (%esp) movl a, %eax testl %eax, %eax js .LBB1_2 #bb .LBB1_1: #cond_next movl b, %ecx movl (%ecx), %ecx cmpl %eax, %ecx jge .LBB1_3 #UnifiedReturnBlock .LBB1_2: #bb movl b, %eax movl $0, (%eax) xorl %eax, %eax addl $2, %esp ret .LBB1_3: #UnifiedReturnBlock xorl %eax, %eax addl $2, %esp ret -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 20:08:14 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 20:08:14 -0500 Subject: [LLVMbugs] [Bug 1309] [meta] DSE is not aggressive enough Message-ID: <200704250108.l3P18E4J020402@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1309 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | Summary|DSE is not aggressive enough|[meta] DSE is not aggressive | |enough ------- Additional Comments From sabre at nondot.org 2007-04-24 20:08 ------- Okay, sorry, I guess this should be the meta bug for relating DSE improvements. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Tue Apr 24 20:17:33 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Tue, 24 Apr 2007 20:17:33 -0500 Subject: [LLVMbugs] [Bug 1354] NEW: Avoid compiling insn-attrtab.c etc. to build llvm-gcc Message-ID: <200704250117.l3P1HXRt020657@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1354 Summary: Avoid compiling insn-attrtab.c etc. to build llvm-gcc Product: Build scripts Version: cvs Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Makefiles AssignedTo: unassignedbugs at nondot.org ReportedBy: evan.cheng at apple.com We should not be compiling insn-attrtab.c and friends. These can take a long time to build, especially if --enable-checking is specified. Would someone who knows about gcc build system please take a look? :-) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 00:04:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 00:04:22 -0500 Subject: [LLVMbugs] [Bug 1355] NEW: TestRunner enhancement request Message-ID: <200704250504.l3P54MVv025134@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1355 Summary: TestRunner enhancement request Product: Test Suite Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org The new TestRunner is much nicer than the old one, and very handy. One (hopefully easy) feature request: It would be nice for TestRunner to take multiple tests on the command line. I usually do: cd test/whatever; find . -name \*.ll -print -exec ../../TestRunner.sh {} \; It would be much nicer to do: cd test/whatever; TestRunner * -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 01:45:11 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 01:45:11 -0500 Subject: [LLVMbugs] [Bug 1356] NEW: Incorrect code generated for function consisting only of inline asm Message-ID: <200704250645.l3P6jBdc026887@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1356 Summary: Incorrect code generated for function consisting only of inline asm Product: new-bugs Version: unspecified Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: major Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: oliver at apple.com llvm-gcc generates invalid code for a function consisting only of a single inline asm statement ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 04:26:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 04:26:16 -0500 Subject: [LLVMbugs] [Bug 1357] NEW: Miscompilation During Compilation of GMP Message-ID: <200704250926.l3P9QGPT007180@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1357 Summary: Miscompilation During Compilation of GMP Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: major Priority: P2 Component: Backend: PowerPC AssignedTo: unassignedbugs at nondot.org ReportedBy: isanbard at gmail.com I've attached the .ll file that aborts when run. It's something to do with the inline assembly thingy. On the plus side, GMP compiled! :-) ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 09:31:29 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 09:31:29 -0500 Subject: [LLVMbugs] [Bug 1017] LLVM should support symbol aliasing Message-ID: <200704251431.l3PEVT0s012739@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1017 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-04-25 09:31 ------- Implemented with: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048406.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048407.html http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048408.html ... http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048418.html llvm-gcc part will follow ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 09:36:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 09:36:16 -0500 Subject: [LLVMbugs] [Bug 1358] NEW: miscompilation of 2007-04-25-weak.c Message-ID: <200704251436.l3PEaGGF012869@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1358 Summary: miscompilation of 2007-04-25-weak.c Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: lauro.venancio at gmail.com LLVM is considering that the if is always true. test: extern int test_weak () __attribute__ ((weak)); int main(){ int (*t)() = test_weak; if (t) return t(); else return 250; } llc emits: .text .align 16 .globl main .type main, at function main: subl $4, %esp fnstcw 2(%esp) movb $2, 3(%esp) fldcw 2(%esp) .LBB1_1: #cond_true call test_weak addl $4, %esp ret .size main, .-main .weak test_weak gcc emits: .file "teste.c" .text .p2align 4,,15 .globl main .type main, @function main: leal 4(%esp), %ecx andl $-16, %esp pushl -4(%ecx) movl $test_weak, %eax movl $250, %edx pushl %ebp movl %esp, %ebp pushl %ecx subl $4, %esp testl %eax, %eax je .L4 call test_weak movl %eax, %edx .L4: addl $4, %esp movl %edx, %eax popl %ecx popl %ebp leal -4(%ecx), %esp ret .size main, .-main .weak test_weak .ident "GCC: (GNU) 4.1.2 20060928 (prerelease) (Ubuntu 4.1.1-13ubuntu5)" .section .note.GNU-stack,"", at progbits ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 09:40:42 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 09:40:42 -0500 Subject: [LLVMbugs] [Bug 1006] llvm-gcc requires --disable-shared on linux Message-ID: <200704251440.l3PEegMX012973@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1006 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-04-25 09:40 ------- Fixed with http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048419.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 09:40:51 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 09:40:51 -0500 Subject: [LLVMbugs] [Bug 843] llvm-gcc4: libstdc++ missing LTHUNKs Message-ID: <200704251440.l3PEepIF012997@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=843 asl at math.spbu.ru changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From asl at math.spbu.ru 2007-04-25 09:40 ------- Fixed with http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048419.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 15:37:25 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 15:37:25 -0500 Subject: [LLVMbugs] [Bug 1359] NEW: Finish Danny Berlin's Re-implementation of Andersen's AA Message-ID: <200704252037.l3PKbPQ7020773@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1359 Summary: Finish Danny Berlin's Re-implementation of Andersen's AA Product: libraries Version: trunk Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: Global Analyses AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at reidspencer.com >From Danny: Hi guys, i'm not going to have time to work on this for a month or two, so i figured i'd post it. This patch 1. reworks the andersen's implementation to support field sensitivity (though field-sensitive constraints are not generated directly yet), and uses it to do indirect function call support. 2. Rewrites the solver to be state of the art in terms of speed. kimwitu++ used to take to solve, and now it takes < 2 seconds. 3. Reworks the rest of the implementation to make it easy to support offline variable substitution. There are still more improvements to be made, and in particular implementing ovs. This is not more than a couple hundred lines of code, and would speed it up by another few orders of magnitude (as well as reduce memory usage greatly). The main thing blocking this patch, however, is that someone needs to rewrite bitmap.c/bitmap.h, obstack.c and obstack.h, into C++. They are currently just modified versions of what gcc is using. You can get rid of the obstacks, of course. Using set or BitVector or something like it will result in a slowdown of mammoth proporations, and about a 10x memory increase. Trust me here, you don't want to use a non-sparse bitmap. BDD's are acceptable, but are about a 2x slowdown. With various optimizations that are going to be published in the next few months, you can get bitmaps to use just as little memory as BDDs would, while being much faster. Anyway, i've attached the patch. The two new .h files go include/llvm/ADT the two new .cc files go into lib/VMCore Of course, you can put them wherever you really want. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 20:20:21 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 20:20:21 -0500 Subject: [LLVMbugs] [Bug 1348] x86_64 codegen bug in md5.c Message-ID: <200704260120.l3Q1KLHJ026405@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1348 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 21:00:35 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 21:00:35 -0500 Subject: [LLVMbugs] [Bug 1360] NEW: MiBench/consumer-typeset failing on Mac OS X Message-ID: <200704260200.l3Q20Z2Y027201@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1360 Summary: MiBench/consumer-typeset failing on Mac OS X Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: major Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: evan.cheng at apple.com It's been failing on both X86 and PPC since 2007/4/6 on both X86 and PPC. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 21:18:15 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 21:18:15 -0500 Subject: [LLVMbugs] [Bug 1360] MiBench/consumer-typeset failing on Mac OS X Message-ID: <200704260218.l3Q2IFjQ027522@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1360 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From sabre at nondot.org 2007-04-25 21:18 ------- *** This bug has been marked as a duplicate of 1295 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Wed Apr 25 23:00:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Wed, 25 Apr 2007 23:00:22 -0500 Subject: [LLVMbugs] [Bug 1355] TestRunner enhancement request Message-ID: <200704260400.l3Q40Mjc029658@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1355 rspencer at reidspencer.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From rspencer at reidspencer.com 2007-04-25 23:00 ------- TestRunner.sh has been augmented to process a list of files, each of which it will execute. Note that the example is much more easily done with: make check TESTSUITE=test/whatever Fixed with this patch: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048480.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 26 00:26:02 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 26 Apr 2007 00:26:02 -0500 Subject: [LLVMbugs] [Bug 1355] TestRunner enhancement request Message-ID: <200704260526.l3Q5Q208031455@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1355 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|FIXED | ------- Additional Comments From sabre at nondot.org 2007-04-26 00:26 ------- Not working: $ cd llvm/test/Assembler $ ../TestRunner.sh *.ll *.llx ... couldn't open "/Users/sabre/cvs/llvm/test/2005-02-09-AsmWriterStoreBug.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2005-05-05-OpaqueUndefValues.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2005-12-21-ZeroInitVector.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2006-05-26-VarargsCallEncode.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2006-09-28-CrashOnInvalid.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2006-12-09-Cast-To-Bool.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2007-01-01-Cast-To-Bool.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2007-01-02-Undefined-Arg-Type.ll": no such file or directory couldn't open "/Users/sabre/cvs/llvm/test/2007-01-05-Cmp-ConstExpr.ll": no such file or directory ... ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 26 00:31:23 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 26 Apr 2007 00:31:23 -0500 Subject: [LLVMbugs] [Bug 1358] LLVM assembler incorrectly folds comparison Message-ID: <200704260531.l3Q5VNnO031718@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1358 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|new bugs |LLVM assembly language | |parser Keywords| |miscompilation Product|new-bugs |libraries Resolution| |FIXED Summary|miscompilation of 2007-04- |LLVM assembler incorrectly |25-weak.c |folds comparison Target Milestone|--- |2.0 Version|unspecified |1.9 ------- Additional Comments From sabre at nondot.org 2007-04-26 00:31 ------- Fixed, testcase here: test/Assembler/2007-04-25-AssemblerFoldExternWeak.ll Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048483.html -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 26 12:49:44 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 26 Apr 2007 12:49:44 -0500 Subject: [LLVMbugs] [Bug 1361] NEW: LLVM crashes when a ModulePass requires a FunctionPass Message-ID: <200704261749.l3QHniYK028750@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1361 Summary: LLVM crashes when a ModulePass requires a FunctionPass Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: major Priority: P2 Component: Core LLVM classes AssignedTo: dpatel at apple.com ReportedBy: zhiruz at gmail.com I tried the ModulePass requiring FunctionPass interface but ran into segmentation fault. Basically I changed the HelloWorld sample pass to be a ModulePass and tried to use the LoopInfo pass inside the runOnModule routine. See below for the source code and error messages. //=========Modified HelloWorld Pass===============// struct Hello3 : public ModulePass { virtual bool runOnModule(Module &M) { HelloCounter++; for (Module::iterator i = M.begin(), e = M.end(); i != e; ++i) { LoopInfo &LI = getAnalysis(*i); std::string fname = i->getName(); EscapeString(fname); cerr << "Hello: " << fname << "\n"; } return false; } // We don't modify the program, so we preserve all analyses virtual void getAnalysisUsage(AnalysisUsage &AU) const { AU.setPreservesAll(); AU.addRequired(); }; }; RegisterPass Z("hello3", "Hello World Pass (ModulePass requires FunctionPass)"); //=========== Error message ================// opt -load Hello.so old.bc -hello3 -o new.bc opt((anonymous namespace)::PrintStackTrace()+0x1a)[0x86423ca] opt((anonymous namespace)::SignalHandler(int)+0x110)[0x86426f2] /lib/tls/libc.so.6[0x683898] opt(std::vector >::begin()+0x19)[0x84e3bb7] opt(llvm::PMTopLevelManager::findAnalysisPass(llvm::PassInfo const*)+0x26)[0x85cc88a] opt(llvm::PMDataManager::findAnalysisPass(llvm::PassInfo const*, bool)+0xb7)[0x85cc827] opt(llvm::PMDataManager::initializeAnalysisImpl(llvm::Pass*)+0x96)[0x85cd39e] opt(llvm::FPPassManager::runOnFunction(llvm::Function&)+0xff)[0x85ce667] opt(llvm::MPPassManager::getOnTheFlyPass(llvm::Pass*, llvm::PassInfo const*, llvm::Function&)+0x62)[0x85ce85e] opt(llvm::AnalysisResolver::findImplPass(llvm::Pass*, llvm::PassInfo const*, llvm::Function&)+0x33)[0x85cbc39] /home/zhiruz/AutoESL/AutoPilot/trunk/build/obj-Linux-x86-g++/Debug/lib/ libLLVMHello.so(llvm::LoopInfo& llvm::Pass::getAnalysisID(llvm::PassInfo const*, llvm::Function&)+0x99)[0x12efc7] /home/zhiruz/AutoESL/AutoPilot/trunk/build/obj-Linux-x86-g++/Debug/lib/ libLLVMHello.so(llvm::LoopInfo& llvm::Pass::getAnalysis(llvm::Function&)+0x63)[0x12f0dd] /home/zhiruz/AutoESL/AutoPilot/trunk/build/obj-Linux-x86-g++/Debug/lib/ libLLVMHello.so((anonymous namespace)::Hello3::runOnModule(llvm::Module&)+0x7e)[0x12f162] opt(llvm::MPPassManager::runOnModule(llvm::Module&)+0x11d)[0x85ce341] opt(llvm::PassManagerImpl::run(llvm::Module&)+0x6e)[0x85ce50e] opt(llvm::PassManager::run(llvm::Module&)+0x1a)[0x85ce562] opt(main+0x975)[0x8371791] /lib/tls/libc.so.6(__libc_start_main+0xd3)[0x670de3] opt(__gxx_personality_v0+0x14d)[0x8364041] Segmentation fault ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 26 12:52:27 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 26 Apr 2007 12:52:27 -0500 Subject: [LLVMbugs] [Bug 1361] LLVM crashes when a ModulePass requires a FunctionPass Message-ID: <200704261752.l3QHqRY5028863@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1361 dpatel at apple.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From dpatel at apple.com 2007-04-26 12:52 ------- Fixed :) http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048503.html ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 26 14:28:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 26 Apr 2007 14:28:17 -0500 Subject: [LLVMbugs] [Bug 1362] NEW: alignment on malloc instruction is not codegened correctly Message-ID: <200704261928.l3QJSHo2030855@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1362 Summary: alignment on malloc instruction is not codegened correctly Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: alenhar2 at uiuc.edu the backends (including C and x86) do not respect alignment on malloc. consider test/Features/alignment.ll. The final malloc (%T) should be codegened to a memalign to preserve the alignment. Once this is done, all memalign and posix_memalign calls can be safely transformed into llvm malloc instructions. Plus it is just broken right now. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Thu Apr 26 16:09:15 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Thu, 26 Apr 2007 16:09:15 -0500 Subject: [LLVMbugs] [Bug 1339] [x86] dollar-signs in identifiers in assembly not protected Message-ID: <200704262109.l3QL9FtS000667@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1339 djg at cray.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED ------- Additional Comments From djg at cray.com 2007-04-26 16:09 ------- Fixed. Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048526.html Testcase here: test/CodeGen/X86/dollar-name.ll ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 27 12:13:44 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 27 Apr 2007 12:13:44 -0500 Subject: [LLVMbugs] [Bug 1326] llc -enable-eh crash Message-ID: <200704271713.l3RHDiJx032383@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1326 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|new bugs |Common Code Generator Code Keywords| |compile-fail Product|new-bugs |libraries Resolution| |FIXED Target Milestone|--- |2.0 Version|unspecified |trunk ------- Additional Comments From sabre at nondot.org 2007-04-27 12:13 ------- Fixed, testcase here: CodeGen/Generic/2007-04-14-EHSelectorCrash.ll Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048557.html -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 27 15:49:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 27 Apr 2007 15:49:22 -0500 Subject: [LLVMbugs] [Bug 1363] NEW: protected visibility not supported Message-ID: <200704272049.l3RKnM6u004493@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1363 Summary: protected visibility not supported Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: lauro.venancio at gmail.com LLVM doesn't support int bar __attribute__ ((visibility ("protected"))) = 1; This is needed to compile the glibc. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Fri Apr 27 19:58:38 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Fri, 27 Apr 2007 19:58:38 -0500 Subject: [LLVMbugs] [Bug 1345] instcombine should recover more type info Message-ID: <200704280058.l3S0wccC010047@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1345 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED Target Milestone|--- |2.0 Version|trunk |1.0 ------- Additional Comments From sabre at nondot.org 2007-04-27 19:58 ------- Implemented, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048573.html Testcase here: Transforms/InstCombine/bitcast-gep.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 00:27:31 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 00:27:31 -0500 Subject: [LLVMbugs] [Bug 1364] NEW: Assertion failure in PassManager Message-ID: <200704280527.l3S5RVBa014745@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1364 Summary: Assertion failure in PassManager Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: babic.domagoj at gmail.com The following piece of code: #include "llvm/Pass.h" #include "llvm/Analysis/Dominators.h" #include "llvm/Analysis/PostDominators.h" using namespace llvm; class PathExpressions : public FunctionPass { public: virtual void getAnalsisUsage(AnalysisUsage &AU) const { AU.addRequired(); AU.addRequired(); } virtual bool runOnFunction(Function &F) { DominatorTree &DT = getAnalysis(); PostDominatorTree &PDT = getAnalysis(); return false; } }; namespace { RegisterPass P("pathexpr", "Compute path expressions."); } // End of anonymous namespace ----- when compiled as a loadable module triggers an assertion: opt -load=out.so -disable-output -disable-verify -debug-pass=Details -pathexpr anybc.bc Pass Arguments: -pathexpr Target Data Layout ModulePass Manager FunctionPass Manager Compute path expressions. -- Compute path expressions. 0x114f2b0 Executing Pass 'Function Pass Manager' on Module 'anybc.bc... 0xe32bf0 Executing Pass 'Compute path expressions.' on Function 'control_printf... opt: /work/projects/llvm/include/llvm/PassAnalysisSupport.h:191: AnalysisType& llvm::Pass::getAnalysisID(const llvm::PassInfo*) const [with AnalysisType = llvm::DominatorTree]: Assertion `ResultPass && "getAnalysis*() called on an analysis that was not " "'required' by pass!"' failed. opt((anonymous namespace)::PrintStackTrace()+0x17)[0xa20ba7] opt((anonymous namespace)::SignalHandler(int)+0x10e)[0xa20e76] Domagoj ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 00:32:51 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 00:32:51 -0500 Subject: [LLVMbugs] [Bug 1365] NEW: Assertion failure in PassManager Message-ID: <200704280532.l3S5WpMa014850@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1365 Summary: Assertion failure in PassManager Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: babic.domagoj at gmail.com The following piece of code: #include "llvm/Pass.h" #include "llvm/Analysis/Dominators.h" #include "llvm/Analysis/PostDominators.h" using namespace llvm; class PathExpressions : public FunctionPass { public: virtual void getAnalsisUsage(AnalysisUsage &AU) const { AU.addRequired(); AU.addRequired(); } virtual bool runOnFunction(Function &F) { DominatorTree &DT = getAnalysis(); PostDominatorTree &PDT = getAnalysis(); return false; } }; namespace { RegisterPass P("pathexpr", "Compute path expressions."); } // End of anonymous namespace ----- when compiled as a loadable module triggers an assertion: opt -load=out.so -disable-output -disable-verify -debug-pass=Details -pathexpr anybc.bc Pass Arguments: -pathexpr Target Data Layout ModulePass Manager FunctionPass Manager Compute path expressions. -- Compute path expressions. 0x114f2b0 Executing Pass 'Function Pass Manager' on Module 'anybc.bc... 0xe32bf0 Executing Pass 'Compute path expressions.' on Function 'control_printf... opt: /work/projects/llvm/include/llvm/PassAnalysisSupport.h:191: AnalysisType& llvm::Pass::getAnalysisID(const llvm::PassInfo*) const [with AnalysisType = llvm::DominatorTree]: Assertion `ResultPass && "getAnalysis*() called on an analysis that was not " "'required' by pass!"' failed. opt((anonymous namespace)::PrintStackTrace()+0x17)[0xa20ba7] opt((anonymous namespace)::SignalHandler(int)+0x10e)[0xa20e76] Domagoj ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 00:33:17 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 00:33:17 -0500 Subject: [LLVMbugs] [Bug 1366] NEW: Assertion failure in PassManager Message-ID: <200704280533.l3S5XHtN014878@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1366 Summary: Assertion failure in PassManager Product: libraries Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Core LLVM classes AssignedTo: unassignedbugs at nondot.org ReportedBy: babic.domagoj at gmail.com The following piece of code: #include "llvm/Pass.h" #include "llvm/Analysis/Dominators.h" #include "llvm/Analysis/PostDominators.h" using namespace llvm; class PathExpressions : public FunctionPass { public: virtual void getAnalsisUsage(AnalysisUsage &AU) const { AU.addRequired(); AU.addRequired(); } virtual bool runOnFunction(Function &F) { DominatorTree &DT = getAnalysis(); PostDominatorTree &PDT = getAnalysis(); return false; } }; namespace { RegisterPass P("pathexpr", "Compute path expressions."); } // End of anonymous namespace ----- when compiled as a loadable module triggers an assertion: opt -load=out.so -disable-output -disable-verify -debug-pass=Details -pathexpr anybc.bc Pass Arguments: -pathexpr Target Data Layout ModulePass Manager FunctionPass Manager Compute path expressions. -- Compute path expressions. 0x114f2b0 Executing Pass 'Function Pass Manager' on Module 'anybc.bc... 0xe32bf0 Executing Pass 'Compute path expressions.' on Function 'control_printf... opt: /work/projects/llvm/include/llvm/PassAnalysisSupport.h:191: AnalysisType& llvm::Pass::getAnalysisID(const llvm::PassInfo*) const [with AnalysisType = llvm::DominatorTree]: Assertion `ResultPass && "getAnalysis*() called on an analysis that was not " "'required' by pass!"' failed. opt((anonymous namespace)::PrintStackTrace()+0x17)[0xa20ba7] opt((anonymous namespace)::SignalHandler(int)+0x10e)[0xa20e76] Domagoj ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 00:36:55 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 00:36:55 -0500 Subject: [LLVMbugs] [Bug 1365] Assertion failure in PassManager Message-ID: <200704280536.l3S5at9A014980@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1365 babic.domagoj at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From babic.domagoj at gmail.com 2007-04-28 00:36 ------- Refresh re-submits the bug. Sorry. *** This bug has been marked as a duplicate of 1364 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 00:37:29 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 00:37:29 -0500 Subject: [LLVMbugs] [Bug 1366] Assertion failure in PassManager Message-ID: <200704280537.l3S5bThW015019@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1366 babic.domagoj at gmail.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From babic.domagoj at gmail.com 2007-04-28 00:37 ------- Sorry, refresh problem again... *** This bug has been marked as a duplicate of 1364 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 01:44:22 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 01:44:22 -0500 Subject: [LLVMbugs] [Bug 1356] [inlineasm] integer mem input uses int operand as address instead of value Message-ID: <200704280644.l3S6iMSh016799@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1356 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|new bugs |Common Code Generator Code Keywords| |miscompilation Product|new-bugs |libraries Resolution| |FIXED Summary|Incorrect code generated for|[inlineasm] integer mem |function consisting only of |input uses int operand as |inline asm |address instead of value Target Milestone|--- |2.0 Version|unspecified |1.0 ------- Additional Comments From sabre at nondot.org 2007-04-28 01:44 ------- Fixed. Many patches made this possible, but the final patch is here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070423/048601.html Testcase here: CodeGen/X86/2007-04-27-InlineAsm-IntMemInput.ll Many thanks for the reduced testcase! -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 01:46:03 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 01:46:03 -0500 Subject: [LLVMbugs] [Bug 644] [llvm-g++] "invoke" instructions generated when using -fno-exceptions Message-ID: <200704280646.l3S6k3lA016868@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=644 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|REOPENED |RESOLVED Resolution| |FIXED ------- Additional Comments From sabre at nondot.org 2007-04-28 01:46 ------- This is fixed in llvm-gcc4 and llvm-gcc3 is dead, despite llvm-gcc4 having incomplete EH support. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 07:24:30 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 07:24:30 -0500 Subject: [LLVMbugs] [Bug 1367] NEW: NightlyTester should draw confidence bands for longterm results Message-ID: <200704281224.l3SCOUZH031803@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1367 Summary: NightlyTester should draw confidence bands for longterm results Product: Test Suite Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: enhancement Priority: P2 Component: Nightly Tester AssignedTo: unassignedbugs at nondot.org ReportedBy: asl at math.spbu.ru Some tests shows highly unstable results from run to run. It will be nice to have some sane confidence borders. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sat Apr 28 23:04:45 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sat, 28 Apr 2007 23:04:45 -0500 Subject: [LLVMbugs] [Bug 1368] NEW: Per-changeset build and regression testing Message-ID: <200704290404.l3T44jad014766@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1368 Summary: Per-changeset build and regression testing Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: christopher.lamb at gmail.com As part of SVN integration automatic build regressions should be performed per changeset. This will allow identification of any commits that break the build or the regression suite. The tool that's chosen should provide for automatic notification via email of broken builds. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 29 02:05:16 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 29 Apr 2007 02:05:16 -0500 Subject: [LLVMbugs] [Bug 1369] NEW: [libsystem] Implement better support for new MemoryBuffer class Message-ID: <200704290705.l3T75Giu018078@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1369 Summary: [libsystem] Implement better support for new MemoryBuffer class Product: libraries Version: 1.0 Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: System Library AssignedTo: unassignedbugs at nondot.org ReportedBy: sabre at nondot.org The new MemoryBuffer class (soon to be used by multiple llvm clients) has two major FIXMEs, caused by deficiencies in the libsystem interface: 1. We need a way to portably read a file into a pre-allocated memory buffer for the MemoryBuffer::getFile method. This is easy on Unix (code is provided) but needs a Win32 implementation. 2. MemoryBufferMMapFile ctor: the caller of the ctor knows the length of the file being opened (in fact, it has a fully populated PathWithStatus object). However, the MappedFile::open method takes a path, not a PathWithStatus object, which it then uses to re-stat the file to get its length (an extra, inefficient, syscall). MappedFile::open should just take a PathWithStatus object? -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 29 22:31:48 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 29 Apr 2007 22:31:48 -0500 Subject: [LLVMbugs] [Bug 1370] NEW: Make generic CodeGen tests truly generic Message-ID: <200704300331.l3U3VmSi029437@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1370 Summary: Make generic CodeGen tests truly generic Product: Test Suite Version: 1.0 Platform: All OS/Version: All Status: NEW Severity: minor Priority: P2 Component: DejaGNU AssignedTo: unassignedbugs at nondot.org ReportedBy: rspencer at reidspencer.com The DejaGnu tests will now avoid running tests for targets that are not configured. Unfortunately, the Generic test suite contains test cases for specific targets. This causes false FAIL in test/CodeGen/Generic because the target specific options are not supported by llc. For example, the -march=ppc32 option is not supported in a test case if PowerPC is not configured to be built. This should be rectified by moving or copying the tests that use the -march option from Generic to the appropriate target directories. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Sun Apr 29 22:46:34 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Sun, 29 Apr 2007 22:46:34 -0500 Subject: [LLVMbugs] [Bug 1371] NEW: Cast from unsupported vector assertion from LegalizeDAG Message-ID: <200704300346.l3U3kYiN029888@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1371 Summary: Cast from unsupported vector assertion from LegalizeDAG Product: new-bugs Version: unspecified Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: me22.ca at gmail.com I get the following failure llc: LegalizeDAG.cpp:5671: llvm::SDOperand::SelectionDAGLegalize::PackVectorOp(llvm::SDOperand, llvm::MVT::ValueType): Assertion `0 && "Cast from unsupported vector type not implemented yet!"' failed. from both CodeGen/Generic/vector.ll and CodeGen/X86/2006-10-13-CycleInDAG.ll Full output: Running /home/me22/programming/llvm-cvs/test/CodeGen/Generic/dg.exp ... FAIL: /home/me22/programming/llvm-cvs/test/CodeGen/Generic/vector.ll Failed with signal(SIGABRT) at line 1 while running: llvm-upgrade < /home/me22/programming/llvm-cvs/test/CodeGen/Generic/vector.ll | llvm-as | llc .text .align 16 .globl test_f1 .type test_f1, at function test_f1: movl 4(%esp), %eax flds (%eax) movl 8(%esp), %eax fadds (%eax) movl 12(%esp), %eax fstps (%eax) #FP_REG_KILL ret .size test_f1, .-test_f1 .align 16 .globl test_f2 .type test_f2, at function test_f2: movl 4(%esp), %eax flds 4(%eax) movl 8(%esp), %ecx fadds 4(%ecx) flds (%eax) fadds (%ecx) movl 12(%esp), %eax fxch %st(1) fstps 4(%eax) fstps (%eax) #FP_REG_KILL ret .size test_f2, .-test_f2 .align 16 .globl test_f4 .type test_f4, at function test_f4: movl 4(%esp), %eax movaps (%eax), %xmm0 movl 8(%esp), %eax addps (%eax), %xmm0 movl 12(%esp), %eax movaps %xmm0, (%eax) ret .size test_f4, .-test_f4 .align 16 .globl test_f8 .type test_f8, at function test_f8: movl 4(%esp), %eax movaps 16(%eax), %xmm0 movl 8(%esp), %ecx addps 16(%ecx), %xmm0 movaps (%eax), %xmm1 addps (%ecx), %xmm1 movl 12(%esp), %eax movaps %xmm0, 16(%eax) movaps %xmm1, (%eax) ret .size test_f8, .-test_f8 .align 16 .globl test_fmul .type test_fmul, at function test_fmul: movl 4(%esp), %eax movaps 16(%eax), %xmm0 movl 8(%esp), %ecx mulps 16(%ecx), %xmm0 movaps (%eax), %xmm1 mulps (%ecx), %xmm1 movl 12(%esp), %eax movaps %xmm0, 16(%eax) movaps %xmm1, (%eax) ret .size test_fmul, .-test_fmul .align 16 .globl test_div .type test_div, at function test_div: movl 4(%esp), %eax movaps 16(%eax), %xmm0 movl 8(%esp), %ecx divps 16(%ecx), %xmm0 movaps (%eax), %xmm1 divps (%ecx), %xmm1 movl 12(%esp), %eax movaps %xmm0, 16(%eax) movaps %xmm1, (%eax) ret .size test_div, .-test_div .align 16 .globl test_cst .type test_cst, at function test_cst: subl $76, %esp movl $1073741824, 32(%esp) movl $1036831949, 48(%esp) movl $1083179008, (%esp) movl $1065353216, 16(%esp) movaps 16(%esp), %xmm0 unpcklps (%esp), %xmm0 movaps 48(%esp), %xmm1 unpcklps 32(%esp), %xmm1 unpcklps %xmm0, %xmm1 movl 80(%esp), %eax addps (%eax), %xmm1 movl 84(%esp), %eax movaps %xmm1, (%eax) addl $76, %esp ret .size test_cst, .-test_cst .align 16 .globl test_zero .type test_zero, at function test_zero: xorps %xmm0, %xmm0 movl 4(%esp), %eax addps (%eax), %xmm0 movl 8(%esp), %eax movaps %xmm0, (%eax) ret .size test_zero, .-test_zero .align 16 .globl test_undef .type test_undef, at function test_undef: #IMPLICIT_DEF %xmm0 movl 8(%esp), %eax movaps %xmm0, (%eax) ret .size test_undef, .-test_undef .align 16 .globl test_constant_insert .type test_constant_insert, at function test_constant_insert: subl $28, %esp movl $1092616192, (%esp) xorps %xmm0, %xmm0 movaps (%esp), %xmm1 movss %xmm1, %xmm0 movl 32(%esp), %eax movaps %xmm0, (%eax) addl $28, %esp ret .size test_constant_insert, .-test_constant_insert .align 16 .globl test_variable_buildvector .type test_variable_buildvector, at function test_variable_buildvector: subl $28, %esp fldl 32(%esp) fstps 24(%esp) flds 24(%esp) fstps (%esp) xorps %xmm0, %xmm0 movaps (%esp), %xmm1 movss %xmm1, %xmm0 movl 40(%esp), %eax movaps %xmm0, (%eax) #FP_REG_KILL addl $28, %esp ret .size test_variable_buildvector, .-test_variable_buildvector .align 16 .globl test_scalar_to_vector .type test_scalar_to_vector, at function test_scalar_to_vector: subl $28, %esp fldl 32(%esp) fstps 24(%esp) flds 24(%esp) fstps (%esp) movl 40(%esp), %eax movaps (%esp), %xmm0 movaps %xmm0, (%eax) #FP_REG_KILL addl $28, %esp ret .size test_scalar_to_vector, .-test_scalar_to_vector .align 16 .globl test_extract_elt .type test_extract_elt, at function test_extract_elt: subl $28, %esp movl 32(%esp), %eax movaps (%eax), %xmm0 movaps %xmm0, (%esp) flds 12(%esp) #FP_REG_KILL addl $28, %esp ret .size test_extract_elt, .-test_extract_elt .align 16 .globl test_extract_elt2 .type test_extract_elt2, at function test_extract_elt2: movl 4(%esp), %eax fldl 24(%eax) #FP_REG_KILL ret .size test_extract_elt2, .-test_extract_elt2 llc: LegalizeDAG.cpp:5671: llvm::SDOperand::SelectionDAGLegalize::PackVectorOp(llvm::SDOperand, llvm::MVT::ValueType): Assertion `0 && "Cast from unsupported vector type not implemented yet!"' failed. llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x84e573a] /lib/libc.so.6(abort+0xfe)[0x401adc8e] /lib/libc.so.6(__assert_fail+0xfb)[0x401a5dcb] llc[0x82ece74] Running /home/me22/programming/llvm-cvs/test/CodeGen/X86/dg.exp ... FAIL: /home/me22/programming/llvm-cvs/test/CodeGen/X86/2006-10-13-CycleInDAG.ll Failed with signal(SIGABRT) at line 1 while running: llvm-upgrade < /home/me22/programming/llvm-cvs/test/CodeGen/X86/2006-10-13-CycleInDAG.ll | llvm-as | llc -march=x86 llc: LegalizeDAG.cpp:5671: llvm::SDOperand::SelectionDAGLegalize::PackVectorOp(llvm::SDOperand, llvm::MVT::ValueType): Assertion `0 && "Cast from unsupported vector type not implemented yet!"' failed. llc((anonymous namespace)::PrintStackTrace()+0x1a)[0x84e573a] /lib/libc.so.6(abort+0xfe)[0x401adc8e] /lib/libc.so.6(__assert_fail+0xfb)[0x401a5dcb] llc[0x82ece74] ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 00:02:19 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 00:02:19 -0500 Subject: [LLVMbugs] [Bug 1370] Make generic CodeGen tests truly generic Message-ID: <200704300502.l3U52Jch031291@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1370 tonic at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |DUPLICATE ------- Additional Comments From tonic at nondot.org 2007-04-30 00:02 ------- *** This bug has been marked as a duplicate of 556 *** ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 00:07:15 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 00:07:15 -0500 Subject: [LLVMbugs] [Bug 556] Make degjagnu select targets to test Message-ID: <200704300507.l3U57FTW031455@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=556 rspencer at reidspencer.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |FIXED Target Milestone|--- |2.0 ------- Additional Comments From rspencer at reidspencer.com 2007-04-30 00:07 ------- This has been fixed a week ago. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 00:09:14 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 00:09:14 -0500 Subject: [LLVMbugs] [Bug 1370] Make generic CodeGen tests truly generic Message-ID: <200704300509.l3U59E1G031507@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1370 rspencer at reidspencer.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |REOPENED Resolution|DUPLICATE | Target Milestone|--- |2.0 ------- Additional Comments From rspencer at reidspencer.com 2007-04-30 00:09 ------- This is not a duplicate of 556. bug 556 had to do with the mechanism to detect that only certain targets are built and to not execute target specific tests if the target hasn't been built. That functionality was implemented a week ago so I just closed 556. This bug is about getting tests in the right directory, a wholly different problem. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 12:25:55 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 12:25:55 -0500 Subject: [LLVMbugs] [Bug 1372] NEW: Adding PR to Dejagnu tests has impacted nightly tester Message-ID: <200704301725.l3UHPt4X021976@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1372 Summary: Adding PR to Dejagnu tests has impacted nightly tester Product: Test Suite Version: trunk Platform: PC OS/Version: Linux Status: NEW Severity: normal Priority: P2 Component: Nightly Tester AssignedTo: unassignedbugs at nondot.org ReportedBy: tonic at nondot.org The nightly tester is not aware of the change of adding PR numbers to dejagnu tests. This basically means that the parsing code use to extract the dejagnu test names is broken (still works but isnt correct). It will also impact diffing results between nights since the file name has changed. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 16:12:51 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 16:12:51 -0500 Subject: [LLVMbugs] [Bug 1357] [inlineasm] earlyclobber modifier is silently ignored Message-ID: <200704302112.l3ULCpi8026407@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1357 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Component|Backend: PowerPC |Common Code Generator Code OS/Version|MacOS X |All Platform|Macintosh |All Resolution| |FIXED Summary|[inlineasm] Miscompilation |[inlineasm] earlyclobber |of earlyclobber modifier |modifier is silently ignored Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-30 16:12 ------- Fixed. Patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070430/048747.html There are several other required patches to go along with this though. Testcase here: CodeGen/PowerPC/2007-04-30-InlineAsmEarlyClobber.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 18:36:14 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 18:36:14 -0500 Subject: [LLVMbugs] [Bug 1228] [eh] branchfolding inf recursion with EH enabled Message-ID: <200704302336.l3UNaERd029597@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1228 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Keywords| |compile-fail OS/Version|Linux |All Platform|PC |All Resolution| |FIXED Summary|Segfault with EH's enabled |[eh] branchfolding inf | |recursion with EH enabled Target Milestone|--- |2.0 ------- Additional Comments From sabre at nondot.org 2007-04-30 18:36 ------- Fixed, patch here: http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20070430/048754.html Testcase here: CodeGen/Generic/2007-04-30-LandingPadBranchFolding.ll -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 19:05:30 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 19:05:30 -0500 Subject: [LLVMbugs] [Bug 1373] NEW: Some method to represent alias information in LLVM would be useful Message-ID: <200705010005.l4105UGH030424@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1373 Summary: Some method to represent alias information in LLVM would be useful Product: new-bugs Version: unspecified Platform: All OS/Version: All Status: NEW Severity: enhancement Priority: P2 Component: new bugs AssignedTo: unassignedbugs at nondot.org ReportedBy: christopher.lamb at gmail.com >From this llvm-dev thread (http://lists.cs.uiuc.edu/pipermail/llvmdev/2007-March/008323.html), Chris Lattner writes: I think the goal should be to capture the most important information. The c99 definition of restrict has some nasty and complex implementation issues. It may not be reasonable or interesting to capture every corner case. A more interesting question is how to represent TBAA information in LLVM :). Here is a sketch of a concrete proposal to get us started: 1. Add a new argument attribute "noalias", which can be applied to pointer formals and pointer function results. 2. Add a new intrinsic "i8* @llvm.noalias(i8*) The semantics of 'noalias' are that they "define a new object". Any pointer derived from it ("it" being the result of a no-alias call result, a formal parameter, or the result of llvm.noalias) is considered to not alias: A. any other objects (e.g. global vars, allocas, functions, etc) B. any other "noalias" objects C. any other formal argument parameters or call results. For these purposes, "pointers derived" means getelementptr and bitcast. Can C99 restrict be mapped onto these semantics? Is "C" really safe? With these semantics, it is clear that 'calloc' could be declared to have a result pointer that is noalias for example, but I'm not sure if it would be safe to map restrict onto it. ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 22:42:56 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 22:42:56 -0500 Subject: [LLVMbugs] [Bug 1374] NEW: [inlineasm] More problems with early clobbers Message-ID: <200705010342.l413gulp001927@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1374 Summary: [inlineasm] More problems with early clobbers Product: libraries Version: trunk Platform: Macintosh OS/Version: MacOS X Status: NEW Severity: normal Priority: P2 Component: Common Code Generator Code AssignedTo: unassignedbugs at nondot.org ReportedBy: isanbard at gmail.com This is the same type of problem we were seeing in PR1357: This program: void bork(void) { static const struct { unsigned long want_dh,want_dl, mh,ml, sh,sl; } data[] = { { 0,0, 1,2, 1,2 }, { 0,0, 0,1, 0,1 } }; int i; unsigned long got_dh, got_dl; for (i = 0; i < (sizeof (data) / sizeof ((data)[0])); i++) { __asm__ ("{sf%I4|subf%I4c} %1,%5,%4\n\t{sfe|subfe} %0,%3,%2" : "=r" (got_dh), "=&r" (got_dl) : "r" (data[i].mh), "r" (data[i].sh), "rI" (data[i].ml), "r" (data[i].sl)); if (got_dh != data[i].want_dh || got_dl != data[i].want_dl) { abort (); } } } Gives this error: check_data wrong at data[0] mh=1 ml=2 sh=1 sl=2 want dh=0 want dl=0 got dh =4294967295 got dl =0 Abort trap Generates code like this: # InlineAsm Start subfc r2,r5,r4 subfe r27,r3,r2 # InlineAsm End stw r2, 60(r1) -bw ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. From bugzilla-daemon at cs.uiuc.edu Mon Apr 30 23:07:54 2007 From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu) Date: Mon, 30 Apr 2007 23:07:54 -0500 Subject: [LLVMbugs] [Bug 1374] [inlineasm] More problems with early clobbers Message-ID: <200705010407.l4147sfF002400@zion.cs.uiuc.edu> http://llvm.org/bugs/show_bug.cgi?id=1374 sabre at nondot.org changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WORKSFORME ------- Additional Comments From sabre at nondot.org 2007-04-30 23:07 ------- works for me. I get: # InlineAsm Start subfc r2,r6,r5 subfe r3,r4,r3 # InlineAsm End Make sure to update your CFE *and* your llvm tree. -Chris ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is.